cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
[official-gcc.git] / gcc / ChangeLog
blob69cfa623350eb345e39b56d14cb87d4655df181c
1 2013-08-23  Jan Hubicka  <jh@suse.cz>
3         * cgraph.c (cgraph_turn_edge_to_speculative):
4         Fix debug output.
6 2013-08-23  Jan Hubicka  <jh@suse.cz>
8         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
9         (tree_decl_with_vis): Add FINAL field.
11 2013-08-23  Jeff Law  <law@redhat.com>
13         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
14         the debugging dump when the expression is fully redundant.
16 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
18         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
19         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
20         * pretty-print.c (pp_formatted_text_data): Likewise.
21         (pp_write_text_to_stream): Likewise.
22         (pp_write_text_as_dot_label_to_stream): Likewise.
23         (pp_append_r): Likewise.
24         (pp_format): Likewise.
25         (pp_flush): Likewise.
26         (pp_clear_output_area): Likewise.
27         (pp_append_text): Likewise.
28         (pp_formatted_text): Likewise.
29         (pp_remaining_character_count_for_line): Likewise.
30         (pp_newline): Likewise.
31         (pp_character): Likewise.
32         (output_buffer::~output_buffer): Define.
33         (pretty_printer::~pretty_printer): Destruct output buffer.
34         * pretty-print.h (output_buffer::~output_buffer): Declare.
35         (pretty_printer::~pretty_printer): Declare virtual.
37 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
39         PR other/57324
40         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
41         HOST_WIDE_INT_M1U): New macros.
42         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
43         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
44         unsigned -1 for lshift.
45         * cse.c (cse_insn): Likewise.
46         * double-int.c (rshift_double, lshift_double): Likewise.
47         * builtins.c (fold_builtin_bitop): Likewise.
48         * combine.c (force_to_mode): Likewise.
49         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
50         * simplify-rtx.c (simplify_const_unary_operation,
51         simplify_const_binary_operation): Likewise.
52         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
53         check_va_list_escapes): Likewise.
54         * rtlanal.c (nonzero_bits1): Likewise.
55         * expmed.c (expand_smod_pow2): Likewise.
56         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
58 2013-08-23  Jan Hubicka  <jh@suse.cz>
60         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
61         as having address taken.
63 2013-08-23  Jan Hubicka  <jh@suse.cz>
65         * ipa-utils.h (method_class_type): Declare.
66         * ipa-devirt.c (method_class_type): Export.
68         * cgraphunit.c (analyze_functions): Do basic devirtualization;
69         do not walk base classes of anonymous types.
71 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
73         PR rtl-optimization/58220
74         PR regression/58221
75         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
76         handle SEQUENCE insns properly.
78 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
80         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
81         definition.
82         (pp_newline_and_indent): Likewise.
83         (pp_separate_with): Likewise.
84         * pretty-print.c (pp_newline_and_flush): Define.
85         (pp_newline_and_indent): Likewise.
86         (pp_separate_with): Likewise.
88 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
90         PR target/58218
91         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
92         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
94 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
96         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
97         * gcc/config/i386/i386.md (*movti_internal): Use
98         predicate to determine if EVEX is needed.
99         (*movsi_internal): Ditto.
100         (*movdf_internal): Ditto.
101         (*movsf_internal): Ditto.
102         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
104 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
106         PR tree-optimization/58209
107         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
108         (find_tail_calls): Give up for pointer result types if m is non-NULL.
109         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
110         emit POINTER_PLUS_EXPR.
111         (create_tailcall_accumulator): For pointer result type accumulate in
112         sizetype type.
114 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
116         * configure.ac: Add backslashes missing from the last change.
117         * configure: Regenerate.
119 2013-08-22  Jan Hubicka  <jh@susue.cz>
121         * ipa.c (function_and_variable_visibility): First remember function
122         was global and then make it local.
124 2013-08-22  Julian Brown  <julian@codesourcery.com>
126         * configure.ac: Add aarch64 to list of arches which use "nop" in
127         debug_line test.
128         * configure: Regenerate.
130 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
132         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
133         gnu_libc_has_function.
134         * config/s390/tpf.h: Likewise.
136 2013-08-22  Jan Hubicka  <jh@susue.cz>
138         * timevar.c (validate_phases): Add cast.
140 2013-08-22  Jan Hubicka  <jh@susue.cz>
142         * timevar.c (validate_phases): Use size_t for memory.
143         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
145 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
147         * pretty-print.h (output_buffer::output_buffer): Declare.
148         (pretty_printer::pretty_printer): Likewise.
149         (pp_construct): Remove.
150         * pretty-print.c (output_buffer::output_buffer): Define.
151         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
152         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
153         (print_gimple_expr): Likewise.
154         (print_gimple_seq): Likewise.
155         (gimple_dump_bb): Likewise.
156         * sched-vis.c (dump_value_slim): Likewise.
157         (dump_insn_slim): Likewise.
158         (dump_rtl_slim): Likewise.
159         (str_pattern_slim): Likewise.
160         * tree-mudflap.c (mf_varname_tree): Likewise.
161         * graph.c (print_graph_cfg): Likewise.
162         (start_graph_dump): Likewise.
163         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
164         placement-new.
165         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
166         pretty printer initialization.
167         * coretypes.h (diagnostic_context): Remove superflous type alias
168         declaration.
169         (pretty_printer): Likewise.  Declare directly as a class.
170         (pretty_print_info): Remove declaration as class.
171         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
172         and pp_clear_output_area.
173         (asan_add_global): Likewise.
175 2013-08-22  Jan Hubicka  <jh@suse.cz>
177         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
178         * ipa-utils.h (update_type_inheritance_graph): Declare.
179         (possible_polymorphic_call_target_p): Declare.
180         (possible_polymorphic_call_target_p): New.
181         * ipa-devirt.c: Update toplevel comments.
182         (cached_polymorphic_call_targets): Move up.
183         (odr_type_d): Move ID down.
184         (polymorphic_type_binfo_p): Update comment.
185         (odr_hasher::remove): Likewise;
186         (get_odr_type): Set anonymous_namespace.
187         (dump_odr_type): Dump it.
188         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
189         (maybe_record_node): Record node in cached_polymorphic_call_targets.
190         (record_binfo): Add comment.
191         (free_polymorphic_call_targets_hash): Do not ICE when cache is not built.
192         (devirt_node_removal_hook): Do not iCE when cache is freed.
193         (possible_polymorphic_call_target_p): New predicate.
194         (update_type_inheritance_graph): New function.
196 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
197             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
198             Sergey Lega  <sergey.s.lega@intel.com>
199             Anna Tikhonova  <anna.tikhonova@intel.com>
200             Ilya Tocar  <ilya.tocar@intel.com>
201             Andrey Turetskiy  <andrey.turetskiy@intel.com>
202             Ilya Verbin  <ilya.verbin@intel.com>
203             Kirill Yukhin  <kirill.yukhin@intel.com>
204             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
206         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
207         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
208         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
209         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
210         (OPTION_MASK_ISA_AVX2_UNSET): Update.
211         (OPTION_MASK_ISA_AVX512F_UNSET): New.
212         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
213         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
214         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
215         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
216         OPT_mavx512pf, OPT_mavx512er cases.
217         * config/i386/constraints.md (v): New constraint.
218         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
219         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
220         (bit_AVX512CD): New.
221         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
222         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
223         * config/i386/i386-c.c (ix86_target_macros_internal):
224         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
225         __AVX512PF__.
226         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
227         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
228         * config/i386/i386.c (regclass_map, dbx_register_map)
229         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
230         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
231         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
232         -mavx512pf options.
233         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
234         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
235         -mavx512pf options.  Fix formatting.
236         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
237         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
238         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
239         -mavx512cd, -mavx512pf options.
240         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
241         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
242         (ix86_preferred_output_reload_class): Replace SSE_REGS with
243         ALL_SSE_REGS.
244         (ix86_hard_regno_mode_ok): Support 512-bit registers.
245         (ix86_set_reg_reg_cost): Ditto.
246         (x86_order_regs_for_local_alloc): Ditto.
247         (MAX_VECT_LEN): Extend to 64-byte.
248         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
249         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
250         (TARGET_AVX512ER, TARGET_AVX512CD): New.
251         (BIGGEST_ALIGNMENT): Extend to 512-bits.
252         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
253         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
254         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
255         (SSE_REG_MODE_P): Support new modes.
256         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
257         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
258         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
259         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
260         (REG_CLASS_CONTENTS): Add new registers.
261         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
262         (EXT_REX_SSE_REGNO_P): New.
263         (HI_REGISTER_NAMES): Add new registers.
264         * config/i386/i386.md: Define constants for new registers.
265         (mode): Add new 512-bit modes.
266         (prefix): Support evex prefix.
267         (isa): Support avx512f, noavx512f, fma_avx512f.
268         (ssemodesuffix): Add new 512-bit modes.
269         (movxi): New.
270         (*movxi_internal_avx512f): Ditto.
271         (*movdi_internal): Replace constraint "x" with the new constraint "v".
272         Support MODE_XI.
273         (*movsi_internal): Likewise.
274         (*movdf_internal): Likewise.
275         (*movsf_internal): Likewise.
276         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
277         (<code><mode>3): Likewise.
278         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd): New.
279         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
280         with the new constraint "v".
281         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
282         modes.
283         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
284         with the new constraint "v".
285         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
286         (<sse2>_storedqu<avxsizesuffix>): Likewise.
287         (*<plusminus_insn><mode>3): Likewise.
288         (<sse>_vm<plusminus_insn><mode>3): Likewise.
289         (*mul<mode>3): Likewise.
290         (<sse>_vmmul<mode>3): Likewise.
291         (<sse>_div<mode>3): Likewise.
292         (<sse>_vmdiv<mode>3): Likewise.
293         (<sse>_sqrt<mode>2): Likewise.
294         (<sse>_vmsqrt<mode>2): Likewise.
295         (*<code><mode>3_finite): Likewise.
296         (*<code><mode>3) <smaxmin>: Likewise.
297         (<sse>_vm<code><mode>3): Likewise.
298         (*<code><mode>3) <any_logic>: Likewise.
299         (*fma_fmadd_<mode>): Likewise.
300         (*fma_fmsub_<mode>): Likewise.
301         (*fma_fnmadd_<mode>): Likewise.
302         (*fma_fnmsub_<mode>): Likewise.
303         (*fma_fmaddsub_<mode>): Likewise.
304         (*fma_fmsubadd_<mode>): Likewise.
305         (*fmai_fmadd_<mode>): Likewise.
306         (*fmai_fmsub_<mode>): Likewise.
307         (*fmai_fnmadd_<mode>): Likewise.
308         (*fmai_fnmsub_<mode>): Likewise.
309         (sse_cvtsi2ss): Likewise.
310         (sse_cvtsi2ssq): Likewise.
311         (sse_cvtss2si): Likewise.
312         (sse_cvtss2si_2): Likewise.
313         (sse_cvtss2siq): Likewise.
314         (sse_cvtss2siq_2): Likewise.
315         (sse_cvttss2si): Likewise.
316         (sse_cvtss2siq_2): Likewise.
317         (float<sseintvecmodelower><mode>2): Likewise.
318         (sse2_cvtsd2si_2): Likewise.
319         (sse2_cvtsd2siq_2): Likewise.
320         (*<plusminus_insn><mode>3): Likewise.
321         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
322         (*<sse4_1_avx2>_mul<mode>3): Likewise.
323         (ashr<mode>3): Likewise.
324         (<shift_insn><mode>3): Likewise.
325         (avx2_<code><mode>3): Likewise.
326         (*avx2_<code><mode>3): Likewise.
327         (*andnot<mode>3): Likewise.
328         (*<code><mode>3) <any_logic>: Likewise.
329         (abs<mode>2): Likewise.
330         (avx2_permvar<mode>): Likewise.
331         (avx2_perm<mode>_1): Likewise.
332         (*avx_vpermilp<mode>): Likewise.
333         (avx_vpermilvar<mode>3): Likewise.
334         (avx2_ashrv<mode>): Likewise.
335         (avx2_<shift_insn>v<mode>): Likewise.
336         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
337         -mavx512cd.
338         * doc/rtl.texi: Document XImode.
340 2013-08-21  Jeff Law  <law@redhat.com>
342         * tree-flow.h (register_jump_thread): Pass vector of edges
343         instead of each important edge.
344         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
345         thread path into a vector and pass that to register_jump_thread.
346         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
347         passed in edge vector to the current 3-edge form.
349         Revert:
350         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
352         * dce.c (fini_dce): Call df_analyze again just in case
353         delete_unmarked_insns removed anything.
355 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
357         * reload.h (struct reg_equivs): Rename to ..
358         (struct reg_equivs_s): .. this.
360 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
362         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
364 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
366         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest
367         comment.
369 2013-08-21  Jeff Law  <law@redhat.com>
371         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
372         simplify assignments too.  If the RHS collapses to a singleton
373         range, then return the value for the range.
375 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
377         * config/i386/sse.md (V16): Rename to...
378         (VMOVE): this.
379         (mov<mode>): Update iterator name.
380         (*mov<mode>_internal): Ditto.
381         (push<mode>1): Ditto.
382         (movmisalign<mode>): Ditto.
384 2013-08-20  Jan Hubicka  <jh@suse.cz>
386         PR bootstrap/58186
387         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
388         entry for direct edges.
389         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
391 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
393         Revert my last two changes, r201865 and r201864:
395         Revert r201865:
396         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
398         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
399         instances can own GC refs.
401         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
402         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
403         (gcc::context::gt_pch_nx): Likewise.
404         (gcc::context::gt_pch_nx):  Likewise.
405         * ggc.h (gt_ggc_mx <T>): New.
406         (gt_pch_nx_with_op <T>): New.
407         (gt_pch_nx <T>): New.
408         * passes.c (opt_pass::gt_ggc_mx): New.
409         (opt_pass::gt_pch_nx): New.
410         (opt_pass::gt_pch_nx_with_op): New.
411         (pass_manager::gt_ggc_mx): New.
412         (pass_manager::gt_pch_nx): New.
413         (pass_manager::gt_pch_nx_with_op): New.
414         (pass_manager::operator new): Use
415         ggc_internal_cleared_alloc_stat rather than xcalloc.
416         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
417         (pass_manager::gt_ggc_mx): New.
418         (pass_manager::gt_pch_nx): New.
419         (pass_manager::gt_pch_nx_with_op): New.
420         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
421         (opt_pass::operator new): New.
422         (opt_pass::gt_ggc_mx): New.
423         (opt_pass::gt_pch_nx): New.
424         (opt_pass::gt_pch_nx_with_op): New.
426         Revert r201864:
427         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
429         * Makefile.in (GTFILES): Add context.h.
430         * context.c (gcc::context::operator new): New.
431         (gcc::context::gt_ggc_mx): New.
432         (gcc::context::gt_pch_nx): New.
433         (gcc::context::gt_pch_nx): New.
434         * context.h (gcc::context): Add GTY((user)) marking.
435         (gcc::context::operator new): New.
436         (gcc::context::gt_ggc_mx): New.
437         (gcc::context::gt_pch_nx): New.
438         (gcc::context::gt_pch_nx): New.
439         (g): Add GTY marking.
440         (gt_ggc_mx (gcc::context *)): New.
441         (gt_pch_nx (gcc::context *)): New.
442         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
443         void *cookie)): New.
444         * gengtype.c (open_base_files) <ifiles>: Add context.h.
446 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
448         * dce.c (fini_dce): Call df_analyze again just in case
449         delete_unmarked_insns removed anything.
451 2013-08-20  Teresa Johnson  <tejohnson@google.com>
453         PR rtl-optimizations/57451
454         * final.c (reemit_insn_block_notes): Prevent lexical blocks
455         from crossing split section boundaries.
457 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
459         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
460         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
461         with MULTLIB_DEFAULTS.
463 2013-08-20  Nick Clifton  <nickc@redhat.com>
465         * target.def (narrow_volatile_bitfield): Note that the default
466         value is false, not !TARGET_STRICT_ALIGN.
467         * doc/tm.texi: Regenerate.
469 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
471         Fix LIB_SPEC for systems without libpthread.
473         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
474         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
475         for Android.
476         * config/i386/linux-common.h: Likewise.
477         * config/mips/linux-common.h: Likewise.
479 2013-08-20  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
481         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
482         checks.
484 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
486         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
487         instances can own GC refs.
489         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
490         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
491         (gcc::context::gt_pch_nx): Likewise.
492         (gcc::context::gt_pch_nx):  Likewise.
493         * ggc.h (gt_ggc_mx <T>): New.
494         (gt_pch_nx_with_op <T>): New.
495         (gt_pch_nx <T>): New.
496         * passes.c (opt_pass::gt_ggc_mx): New.
497         (opt_pass::gt_pch_nx): New.
498         (opt_pass::gt_pch_nx_with_op): New.
499         (pass_manager::gt_ggc_mx): New.
500         (pass_manager::gt_pch_nx): New.
501         (pass_manager::gt_pch_nx_with_op): New.
502         (pass_manager::operator new): Use
503         ggc_internal_cleared_alloc_stat rather than xcalloc.
504         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
505         (pass_manager::gt_ggc_mx): New.
506         (pass_manager::gt_pch_nx): New.
507         (pass_manager::gt_pch_nx_with_op): New.
508         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
509         (opt_pass::operator new): New.
510         (opt_pass::gt_ggc_mx): New.
511         (opt_pass::gt_pch_nx): New.
512         (opt_pass::gt_pch_nx_with_op): New.
514 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
516         * Makefile.in (GTFILES): Add context.h.
517         * context.c (gcc::context::operator new): New.
518         (gcc::context::gt_ggc_mx): New.
519         (gcc::context::gt_pch_nx): New.
520         (gcc::context::gt_pch_nx): New.
521         * context.h (gcc::context): Add GTY((user)) marking.
522         (gcc::context::operator new): New.
523         (gcc::context::gt_ggc_mx): New.
524         (gcc::context::gt_pch_nx): New.
525         (gcc::context::gt_pch_nx): New.
526         (g): Add GTY marking.
527         (gt_ggc_mx (gcc::context *)): New.
528         (gt_pch_nx (gcc::context *)): New.
529         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
530         void *cookie)): New.
531         * gengtype.c (open_base_files) <ifiles>: Add context.h.
533 2013-08-20  Alan Modra  <amodra@gmail.com>
535         PR target/57865
536         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
537         (rs6000_emit_epilogue): Likewise.
539 2013-08-19  Dehao Chen (dehao@google.com)
541         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
543 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
544             Jakub Jelinek  <jakub@redhat.com>
546         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
547         (BUILT_IN_FABSD64): Likewise.
548         (BUILT_IN_FABSD128): Likewise.
549         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
550         (fold_builtin_1): Likewise.
551         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping destination
552         and source operands.
553         (*abstd2_fpr): Likewise.
554         (*nabstd2_fpr): Likewise.
556 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
558         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
559         JUMP_P and INSN_P.
561 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
563         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
564         incomplete.
566 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
568         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
569         * builtins.c (default_libc_has_function): New.
570         (gnu_libc_has_function): Ditto.
571         (no_c99_libc_has_function): Ditto.
572         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
573         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
574         (fold_builtin_sincos): Likewise.
575         (fold_builtin_cexp): Likewise.
576         * builtins.def (DEF_C94_BUILTIN): Likewise.
577         (DEF_C99_BUILTIN): Likewise.
578         (DEF_C99_C90RES_BUILTIN): Likewise.
579         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
580         definitions to using this define.
581         * config/darwin-protos.h (darwin_libc_has_function): New.
582         * config/darwin.c: (darwin_libc_has_function: Ditto.
583         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
584         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
585         * config/darwin.h: Ditto.
586         * config/elfos.h: Ditto.
587         * config/freebsd.h: Ditto.
588         * config/i386/cygming.h: Ditto.
589         * config/i386/djgpp.h: Ditto.
590         * config/i386/i386-interix.h: Ditto.
591         * config/microblaze/microblaze.h: Ditto.
592         * config/mmix/mmix.h: Ditto.
593         * config/gnu-user.h: Ditto.
594         * config/ia64/hpux.h: Ditto.
595         * config/pa/pa-hpux.h: Ditto.
596         * config/pdp11/pdp11.h: Ditto.
597         * config/picochip/picochip.h: Ditto.
598         * config/linux.h: Ditto.
599         * config/netbsd.h: Ditto.
600         * config/openbsd.h: Ditto.
601         * config/rs6000/aix43.h: Ditto.
602         * config/rs6000/aix51.h: Ditto.
603         * config/rs6000/aix52.h: Ditto.
604         * config/rs6000/aix53.h: Ditto.
605         * config/rs6000/aix61.h: Ditto.
606         * config/rs6000/darwin.h: Ditto.
607         * config/rs6000/linux.h: Ditto.
608         * config/rs6000/linux64.h: Ditto.
609         * config/s390/tpf.h: Ditto.
610         * config/sol2-10.h: Ditto.
611         * config/sol2.h: Ditto.
612         * config/vms/vms.h: Ditto.
613         * config/vxworks.h: Ditto.
614         * config/linux-android.c (linux_android_libc_has_function):
615         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
616         * config/linux-protos.h (linux_android_libc_has_function):
617         New declaration.
618         * config/i386/i386.c (ix86_libc_has_function): New.
619         * config/i386/i386-protos.h
620         (ix86_libc_has_function): New declaration.
621         * config/i386/i386.md
622         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
623         ("isinf<mode>2): Likewise.
624         * convert.c (convert_to_integer): Using new target hook
625         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
626         TARGET_C99_FUNCTIONS.
627         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
628         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
629         * coretypes.h (function_class): New enum for different
630         classes of functions.
631         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
632         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
633         (TARGET_HAS_SINCOS): Likewise.
634         (TARGET_LIBC_HAS_FUNCTION): New.
635         * doc/tm.texi: Regenerated.
636         * targhooks.h (default_libc_has_function): New declaration.
637         (no_c99_libc_has_function): Ditto.
638         (gnu_libc_has_function): Ditto.
639         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
640         and TARGET_HAS_SINCOS.
642 2013-08-18  Jan Hubicka  <jh@suse.cz>
644         * Makeifle-in (ipa-devirt.o): New.
645         (GTFILES): Add ipa-utils.h and ipa-devirt.c
646         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
647         (analyze_functions): Look into possible targets of polymorphic call.
648         * dumpfile.c (dump_files): Add type-inheritance dump.
649         * dumpfile.h (TDI_inheritance): New.
650         * ipa-devirt.c: New file.
651         * ipa-utils.h (odr_type_d): Forward declare.
652         (odr_type): New type.
653         (build_type_inheritance_graph): Declare.
654         (possible_polymorphic_call_targets): Declare and introduce inline
655         variant when only edge is pased.
656         (dump_possible_polymorphic_call_targets): Likewise.
657         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
658         * tree.c (type_in_anonymous_namespace_p): Break out from ...
659         (types_same_for_odr): ... here.
660         * tree.h (type_in_anonymous_namespace_p): Declare.
662 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
664         PR tree-optimization/58006
665         * tree-parloops.c (take_address_of): Don't ICE if get_name
666         returns NULL.
667         (eliminate_local_variables_stmt): Remove clobber stmts.
669 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
671         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
672         error message is issued for an alias to undefined symbol.
674 2013-08-18  Jan Hubicka  <jh@suse.cz>
676         * cgraph.c (cgraph_create_indirect_edge): Discover
677         polymorphic calls and record basic info into indirect_info.
678         * gimple-fold.c (gimple_fold_call): When doing BINFO based
679         devirtualization, ignore objc function calls.
680         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
681         call with no parm index info.
682         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
683         * tree.c (virtual_method_call_p): New function.
684         * tree.h (virtual_method_call_p): Declare.
686 2013-08-16  Jan Hubicka  <jh@suse.cz>
688         PR middle-end/58179
689         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
691 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
693         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
694         attribute.
696 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
698         * gengtype.c (type_for_name): Add special-case support for
699         locating types within the "gcc::" namespace.
700         (open_base_files): Emit a "using namespace gcc" directive.
702 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
704         PR target/58160
705         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
706         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
708         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
709         array instead of each individual operand as a separate argument.
710         (emit_fusion_gpr_load): Likewise.
711         (expand_fusion_gpr_load): Add new function declaration.
713         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
714         signature to have the operands passed as an array, instead of as
715         separate arguments.  Allow ZERO_EXTEND to be in the memory
716         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
717         depend on the register live/dead flags when peepholes are run.
718         (expand_fusion_gpr_load): New function to be called from the
719         peephole2 pass, to change the register that addis sets to be the
720         target register.
721         (emit_fusion_gpr_load): Change the calling signature to have the
722         operands passed as an array, instead of as separate arguments.
723         Allow ZERO_EXTEND to be in the memory address, and also
724         SIGN_EXTEND if -mpower8-fusion-sign.
726         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
727         unspec enumeration.
728         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
729         adjust the register addis loads up in the peephole2 pass.  Do not
730         depend on the register live/dead state when the peephole pass is
731         done.
733 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
735         * gengtype.c (create_user_defined_type): Ensure that the kind
736         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
737         declaration is seen before the GTY((user)) marking.
739 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
741         PR target/58105
742         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
744 2013-08-16  Jan Hubicka  <jh@suse.cz>
746         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
747         arugment expected_type.
748         (gimple_fold_call): Use it.
749         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
750         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
751         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
752         (try_make_edge_direct_virtual_call): Likewise.
753         * tree.c (obj_type_ref_class): New.
754         * tree.h (obj_type_ref_class): Use it.
756 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
758         * sched-vis.c (rtl_slim_pp_initialized): Remove.
759         (rtl_slim_pp): Likewise.
760         (init_rtl_slim_pretty_print): Likewise.
761         (dump_value_slim):  Don't call it.  Use local pretty printer.
762         (dump_insn_slim): Likewise.
763         (dump_rtl_slim): Likewise.
764         (str_pattern_slim): Likewise.
765         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
766         Simplify.
768 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
770         PR tree-optimization/58164
771         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
772         walk gimple_goto_dest of GIMPLE_GOTO.
774         PR tree-optimization/58165
775         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
776         bi_call must be the last stmt in a bb, don't split_block, instead
777         use fallthru edge from it and give up if there is none.
778         Release conds vector when returning early.
780 2013-08-14  Xinliang David Li  <davidxl@google.com>
782         * config/i386/i386.c (ix86_option_override_internal):
783         Remove unused variable and field.
785 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
787         PR target/57949
788         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
789         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
790         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
791         and Linux, correct BLKmode alignment when 128-bit alignment is
792         required and compatibility flag is not set.
793         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
794         for zero-size arguments when compatibility flag is not set.
796 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
798         PR tree-optimization/58145
799         * tree-sra.c (build_ref_for_offset): If prev_base has
800         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
802 2013-08-14  Xinliang David Li  <davidxl@google.com>
804         * config/i386/i386.c (ix86_option_override_internal):
805         Fix uninitialized variable error.
807 2013-08-14  Xinliang David Li  <davidxl@google.com>
809         * config/i386/i386.opt: Define two new options.
810         * config/i386/x86-tune.def: Add arch selector field in macros.
811         * config/i386/i386.h: Adjust macro definition.
812         * config/i386/i386.c (ix86_option_override_internal):
813         Refactor the code.
814         (parse_mtune_ctrl_str): New function.
815         (set_ix86_tune_features): New function.
816         (ix86_function_specific_restore): Call the new helper function.
818 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
820         PR rtl-optimization/57662
821         * sel-sched.c (code_motion_process_successors): When the current insn
822         is removed after the recursive traversal, break from the loop.
823         Add comments and debug printouts.
825 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
826             Alexandre Oliva  <aoliva@redhat.com>
828         PR target/58067
829         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
830         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
831         there also UNSPEC_PLTOFF.
833 2013-08-14  Marek Polacek  <polacek@redhat.com>
835         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
836         NUM_CONDITIONS bit positions.
838 2013-08-13  Cary Coutant  <ccoutant@google.com>
840         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
841         (attr_checksum): Hash vector contents instead of pointer.
842         (attr_checksum_ordered): Likewise.
844 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
846         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
847         when Pmode != word_mode.  Add length_address attribute.
848         (sse3_monitor_<mode>): Merge from sse3_monitor and
849         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
850         Pmode != word_mode.  Update insn length attribute.
851         * config/i386/i386.c (ix86_option_override_internal): Update
852         ix86_gen_monitor selection for merged sse3_monitor insn.
854 2013-08-13  Julian Brown  <julian@codesourcery.com>
856         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
857         perform invalid legitimization on greater-than-word-size modes for
858         TARGET_E500_DOUBLE.
860 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
862         * ira.c (setup_class_translate_array): Use aclass instead of cl
863         for classes not fully covered by allocno classes.
865 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
867         PR tree-optimization/57661
868         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
869         * tree-inline.c (tree_function_versioning): Initialize it.
870         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
871         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
872         that is not being copied.
874         PR sanitizer/56417
875         * asan.c (instrument_strlen_call): Fix typo in comment.
876         Use char * type even for the lhs of POINTER_PLUS_EXPR.
878 2013-08-13  Steve Ellcey  <sellcey@mips.com>
880         * config/mips/mips.md (prefetch): Use lw instead of ld on
881         loongson in 32bit mode.
883 2013-08-13  Nick Clifton  <nickc@redhat.com>
885         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
887 2013-08-13  Jan Hubicka  <jh@suse.cz>
889         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
890         introduced edge; fix typo in sanity check.
891         (cgraph_resolve_speculation): Export; improve diagnostic.
892         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
893         speculation at type mismatch.
894         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
895         (cgraph_resolve_speculation): Declare.
896         (symtab_can_be_discarded): New function.
897         * value-prof.c (gimple_ic_transform): Remove actual transform code.
898         * ipa-inline-transform.c (speculation_removed): New global var.
899         (clone_inlined_nodes): See if speculation can be removed.
900         (inline_call): If speculations was removed, we growths may not match.
901         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
902         (speculation_useful_p): New function.
903         (resolve_noninline_speculation): New function.
904         (inline_small_functions): Resolve useless speculations.
905         * ipa-inline.h (speculation_useful_p): Declare
906         * ipa.c (can_replace_by_local_alias): Simplify.
907         (ipa_profile): Produce speculative calls in non-lto, too;
908         add simple cost model; produce local aliases.
910 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
912         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
913         PASS_MANAGER_H.
915 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
917         * config/i386/i386.c (ix86_function_versions): Use error + inform.
919 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
921         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
922         iterator instead of X87MODEF.
924 2013-08-12  Perez Read  <netfirewall@gmail.com>
926         PR target/58132
927         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
928         operand 0 for intel asm alternative.
929         (*movabs<mode>_2): Ditto for operand 1.
931 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
933         * config/aarch64/arm_none.h
934         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
936 2013-08-12  Nick Clifton  <nickc@redhat.com>
938         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
940 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
942         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
943         expand for QI/HImode operand to produce more effictive code for
944         unsigned char(short) --> float(double) conversion.
946 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
948         * doc/invoke.texi: Mention that -ftls-model does not force the final
949         model.
951 2013-08-12  Marek Polacek  <polacek@redhat.com>
952             Marc Glisse  <marc.glisse@inria.fr>
954         PR tree-optimization/57980
955         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
956         when creating -1 constant.
958 2013-08-10  Jan Hubicka  <jh@suse.cz>
960         Workaround binutils PR14342.
961         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
962         (gimple_init_edge_profiler): Likewise.
963         (gimple_gen_ic_func_profiler): Likewise.
965 2013-08-09  Jan Hubicka  <jh@suse.cz>
967         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
969 2013-08-09  Xinliang David Li  <davidxl@google.com>
971         * config/i386/stringop.def: New file.
972         * config/i386/stringop.opt: New file.
973         * config/i386/i386-opts.h: Include stringopt.def.
974         * config/i386/i386.opt: Include stringopt.opt.
975         * config/i386/i386.c (ix86_option_override_internal):
976         Override default size based stringop inline strategies with options.
977         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
978         New function.
980 2013-08-09  Jan Hubicka  <jh@suse.cz>
982         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
984 2013-08-09  Jan Hubicka  <jh@suse.cz>
986         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
987         CGRAPH_FREQ_MAX.
988         (dump_cgraph_node): Dump profile-id.
989         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
990         and common_target_probability.
991         * lto-cgraph.c (lto_output_edge): Stream common targets.
992         (lto_output_node): Stream profile ids.
993         (input_node): Stream profile ids.
994         (input_edge): Stream common targets.
995         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
996         * ipa.c: Include value-prof.h
997         (ipa_profile_generate_summary): Turn indirect call statement histograms
998         into common targets.
999         (ipa_profile): Turn common targets into speculative edges.
1001 2013-08-09  Jan Hubicka  <jh@suse.cz>
1003         * cgraph.h (cgraph_node): Add profile_id.
1004         * value-prof.c (cgraph_node_map): Turn into pointer_map.
1005         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1006         (del_node_map): Update.
1007         (find_func_by_funcdef_no): Replace by ...
1008         (find_func_by_profile_id): ... this one.
1009         (gimple_ic_transform): Do not remove useful histograms when
1010         speculation is not done; dump info when indirect call removal
1011         can happen at LTO.
1012         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1013         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1014         (__gcov_indirect_call_profiler_v2): .. this one.
1015         * profile.h (init_node_map): Update.
1016         * coverage.c (coverage_compute_profile_id): New function.
1017         * coverage.h (coverage_compute_profile_id): Declare.
1018         * tree-profile.c (init_ic_make_global_vars): Make
1019         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
1020         (gimple_init_edge_profiler): Update prototype of
1021         __gcov_indirect_call_profiler.
1022         (gimple_gen_ic_func_profiler): Simplify.
1023         (tree_profiling): Use init_node_map
1025 2013-08-09  Jan Hubicka  <jh@suse.cz>
1027         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1028         non-speculative refs.
1029         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1030         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1031         (cgraph_set_call_stmt): Likewise.
1032         (cgraph_create_edge_1): Fix release checking compilatoin;
1033         clear lto_stmt_uid.
1034         (cgraph_free_edge): Free indirect info.
1035         (cgraph_turn_edge_to_speculative): New function.
1036         (cgraph_speculative_call_info): New function.
1037         (cgraph_make_edge_direct): Return direct edge; handle speculation.
1038         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1039         (dump_cgraph_node): Dump speculation.
1040         (verify_edge_count_and_frequency): Accept speculative edges.
1041         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1042         (verify_cgraph_node): Handle speculation.
1043         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1044         (cgraph_set_call_stmt): Update prototype.
1045         (cgraph_make_edge_direct): Update prototype.
1046         (cgraph_speculative_call_info): Declare.
1047         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1048         to change; update call of ipa_find_references.
1049         * ipa-ref.c (ipa_record_reference): Fix return value; clear
1050         lto_stmt_uid and speculative flags.
1051         (ipa_dump_references): Dump speculation.
1052         (ipa_clone_references): Clone speculative flag.
1053         (ipa_clone_referring): Likewise.
1054         (ipa_clone_ref): New function.
1055         (ipa_find_reference): Look into lto_stmt_uids
1056         (ipa_clear_stmts_in_references): Do not clear speculative calls.
1057         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1058         (ipa_find_reference): Update declaration.
1059         (ipa_clone_ref): Declare.
1060         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1061         stream speculative flag.
1062         (lto_output_ref): Stream statements uids and speculation.
1063         (input_ref): Likewise.
1064         (input_edge): Stream speuclation.
1065         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1066         (cgraph_set_call_stmt_including_clones): Handle speculation.
1067         * ipa-inline.c (heap_edge_removal_hook): New function.
1068         (inline_small_functions): Register it.
1069         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1070         also initialize refs.
1071         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1072         edge to change.
1073         (try_make_edge_direct_simple_call): Likewise.
1074         (try_make_edge_direct_simple_call): Likewise.
1075         (update_indirect_edges_after_inlining): Likewise.
1076         (remove_described_reference): Look proper lto_stmt_uid.
1077         (propagate_controlled_uses): Likewise.
1078         (propagate_controlled_uses): Liekwise.
1079         * tree-inline.c (copy_bb): Copy speculative edges.
1080         (redirect_all_calls): New function.
1081         (copy_cfg_body): Do redirection after loop info is updated.
1082         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1084 2013-08-09  Jan Hubicka  <jh@suse.cz>
1086         * lto-streamer-out.c (output_function): Renumber PHIs.
1087         * lto-streamer-in.c (input_function): Likewise.
1089 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1091         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1092         (get_lane_unsigned): Likewise.
1093         (dup_lane_scalar): Likewise.
1094         (get_lane): enable for VALL.
1095         * config/aarch64/aarch64-simd.md
1096         (aarch64_dup_lane_scalar<mode>): Remove.
1097         (aarch64_get_lane_signed<mode>): Likewise.
1098         (aarch64_get_lane_unsigned<mode>): Likewise.
1099         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1100         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1101         (aarch64_get_lane<mode>): Enable for all vector modes.
1102         (aarch64_get_lanedi): Remove misleading constraints.
1103         * config/aarch64/arm_neon.h
1104         (__aarch64_vget_lane_any): Define.
1105         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1106         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1107         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1108         * config/aarch64/iterators.md (VDQQH): New.
1109         (VDQQHS): Likewise.
1110         (vwcore): Likewise.
1112 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
1114         * configure.ac: Add GAS check for LEON instructions on SPARC.
1115         * configure: Regenerate.
1116         * config.in: Likewise.
1117         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1118         sparc*-*-* block.
1119         * config/sparc/sparc.opt (LEON, LEON3): New masks.
1120         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1121         for LEON or LEON3.
1122         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1123         (AS_LEON_FLAG): New macro.
1124         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1125         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1126         Deal with LEON and LEON3 for the memory model.
1127         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1128         (atomic_compare_and_swap<mode>_1): Likewise.
1129         (*atomic_compare_and_swap<mode>_1): Likewise.
1131 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1133         * config/arm/neon.md (vcond): Fix floating-point vector
1134         comparisons against 0.
1136 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
1138         * lra-constraints.c (emit_spill_move): Remove assert.
1139         (process_alt_operands): Add more debugging
1140         output.  Increase reject for spilling into memory.  Decrease
1141         reject for reloading scratch.
1142         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1144 2013-08-08  Steve Ellcey  <sellcey@mips.com>
1146         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1147         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1148         micromips incompatible.  Add nan2008.
1149         (MULTILIB_DIRNAMES): Add nan2008.
1150         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1151         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1152         and micromips incompatible.  Add nan2008.
1153         (MULTILIB_DIRNAMES): Add nan2008.
1154         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1156 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
1158         PR rtl-optimization/58079
1159         * combine.c (combine_simplify_rtx): Avoid using SUBST if
1160         simplify_comparison has widened a comparison with an integer.
1162 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1164         * config/arm/neon.md (movmisalign<mode>): Disable when we
1165         don't allow unaligned accesses.
1166         (*movmisalign<mode>_neon_store): Likewise.
1167         (*movmisalign<mode>_neon_load): Likewise.
1168         (*movmisalign<mode>_neon_store): Likewise.
1169         (*movmisalign<mode>_neon_load): Likewise.
1171 2013-08-08  Jan Hubicka  <jh@suse.cz>
1173         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1174         (make_pass_rebuild_cgraph_edges): Also clear references.
1175         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1176         * ipa-inline-transform.c (inline_transform): Remove all references
1177         after inlining.
1178         * cgraphunit.c (expand_function): Remove all references after
1179         expansion.
1180         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1181         (ipa_find_reference): Rewrite to iterator.
1182         (remove_stmt_references): Likewise.
1183         (ipa_clear_stmts_in_references): New function.
1184         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1185         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1186         clear references.
1187         * ipa-split.c (split_function): Remove references in split function.
1189 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1191         PR target/57431
1192         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1193         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1195 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1197         PR target/56979
1198         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1199         suggested mode for the assignment isn't compatible with the
1200         registers required.
1202 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1204         PR target/58065
1205         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1207 2013-08-07  Xinliang David Li  <davidxl@google.com>
1209         * config/i386/i386.opt: New option -mtune-ctrl=.
1210         * config/i386/x86-tune.def: New file.
1211         * config/i386/i386.h: include x86-tune.def.
1212         * config/i386/i386.c (ix86_option_override_internal):
1213         Parsing -mtune-ctrl= option and set tune features.
1215 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
1217         PR other/12081
1218         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1219         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1220         to gen_2arg_fn_t.
1222 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1224         * rtl.h (update_alignments): Declare.
1225         * final.c (grow_label_align): New function extracted from...
1226         (shorten_branches): ...here.  Call it.
1227         (update_alignments): New function.
1228         * reorg.c (sibling_labels): New variable.
1229         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
1230         the new label along with it onto the sibling_labels vector.
1231         (fill_simple_delay_slots): Adjust call to get_label_before.
1232         (fill_slots_from_thread): Likewise.
1233         (relax_delay_slots): Likewise.
1234         (make_return_insns): Likewise.
1235         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1237 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1239         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1240         document its semantics.
1241         (diagnostic_report_diagnostic): Adjust accordingly.
1243 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1245         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1246         (sparc_option_override): ...and port to new C++ pass API.
1247         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1249 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
1251         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1253 2013-08-06  Caroline Tice  <cmtice@google.com>
1255         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1256         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1257         * tree-pass.h: Add pass_vtable_verify.
1258         * varasm.c (assemble_variable): Add code to properly set the comdat
1259         section and name for the .vtable_map_vars section.
1260         (assemble_vtyv_preinit_initializer): New function.
1261         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
1262         LINK_ONCE flag.
1263         * output.h: Add function decl for assemble_vtv_preinit_initializer.
1264         * vtable-verify.c: New file.
1265         * vtable-verify.h: New file.
1266         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1267         initialiation levels.
1268         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1269         * passes.def: Insert pass_vtable_verify.
1270         * aclocal.m4: Reorder includes.
1271         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
1272         -fvtv-counts options.
1273         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1274         as appropriate, if -fvtable-verify=... is used.
1275         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1276         -fvtable-verify=... is used.
1277         * Makefile.in (OBJS):  Add vtable-verify.o to list.
1278         (vtable-verify.o): Add new build rule.
1279         (GTFILES): Add vtable-verify.c to list.
1280         * common.opt (fvtable-verify=): New flag.
1281         (vtv_priority): Values for fvtable-verify= flag.
1282         (fvtv-counts): New flag.
1283         (fvtv-debug): New flag.
1284         * tree.h (save_vtable_map_decl): New extern function decl.
1286 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1288         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1289         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1290         (pass_data_rl78_devirt): ...new pass_data instance and...
1291         (make_pass_rl78_devirt): ...new function.
1292         (rl78_asm_file_start): Port pass registration to new C++ API.
1294 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1296         * coretypes.h (rtl_opt_pass): Add.
1297         (gcc::context): Add.
1298         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1299         (epiphany_init): Port to new C++ pass API.
1300         (epiphany_optimize_mode_switching): Likewise.
1301         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
1302         (pass_manager::get_pass_mode_switching): New.
1303         (pass_manager::get_pass_peephole2): New.
1304         * mode-switching.c (pass_mode_switching): Add clone method.
1305         * recog.c (pass_peephole2): Add clone method.
1306         (pass_split_all_insns): Add clone method.
1308 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
1310         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
1311         (mips_option_override): ...here, porting to new C++ API for passes.
1313 2013-08-06  Jan Hubicka  <jh@suse.cz>
1315         * cgraph.c (cgraph_get_body): New function based on lto.c
1316         implementation.
1317         * cgraph.h (cgraph_get_body): Declare.
1318         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
1319         LTO paths.
1320         * cgraphunit.c (expand_function): Get body prior expanding.
1321         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
1322         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
1323         really need.
1324         * passes.c (do_per_function_toporder): Get body.
1325         * tree-inline.c (expand_call_inline): Get body prior inlining it.
1326         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
1328 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1330         PR fortran/57987
1331         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
1332         is not re-finalized.  Rename second parameter to no_collect.
1334 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1336         PR middle-end/58041
1337         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
1338         MEM_REF has proper alignment information.
1340 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
1342         PR other/12081
1343         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
1344         class insn_gen_fn.
1345         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
1346         rtx (*) (rtx, ...) with insn_gen_fn.
1347         * genoutput.c (output_insn_data): Cast gen_? function pointers to
1348         insn_gen_fn::stored_funcptr.  Add initializer braces.
1350 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1352         Rewrite how instances of passes are cloned to remove assumptions
1353         about their sizes (thus allowing pass subclasses to have
1354         additional data fields, albeit non-GC-managed ones at this point).
1356         * passes.c (make_pass_instance): Now that passes have clone
1357         methods, rewrite this function to eliminate XNEW and memcpy
1358         calls that used hardcoded sizes.  Since this function no longer
1359         creates pass instances, rename it to...
1360         (add_pass_instance): ...this.  Document the old way that passes were
1361         numbered and flagged, and rework this function to continue using it.
1362         (next_pass_1): Add an initial_pass argument for use by
1363         add_pass_instance.
1364         (position_pass): When adding multiple instances of a pass, use
1365         the pass's clone method, rather than relying on the XNEW/memcpy
1366         within the former make_pass_instance (now add_pass_instance).
1367         (pass_manager::pass_manager): When invoking next_pass_1, also supply
1368         the initial instance of the current pass within the pass manager.
1370 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1372         This is the automated part of the conversion of passes from C
1373         structs to C++ classes.
1375         Patch autogenerated by refactor_passes.py from
1376         https://github.com/davidmalcolm/gcc-refactoring-scripts
1377         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
1379         * asan.c (pass_asan): Convert from a global struct to a subclass of
1380         gimple_opt_pass along with...
1381         (pass_data_asan): ...new pass_data instance and...
1382         (make_pass_asan): ...new function.
1383         (pass_asan_O0): Convert from a global struct to a subclass of
1384         gimple_opt_pass along with...
1385         (pass_data_asan_O0): ...new pass_data instance and...
1386         (make_pass_asan_O0): ...new function.
1387         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
1388         subclass of rtl_opt_pass along with...
1389         (pass_data_inc_dec): ...new pass_data instance and...
1390         (make_pass_inc_dec): ...new function.
1391         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
1392         a subclass of rtl_opt_pass along with...
1393         (pass_data_reorder_blocks): ...new pass_data instance and...
1394         (make_pass_reorder_blocks): ...new function.
1395         (pass_duplicate_computed_gotos): Convert from a global struct to a
1396         subclass of rtl_opt_pass along with...
1397         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
1398         (make_pass_duplicate_computed_gotos): ...new function.
1399         (pass_partition_blocks): Convert from a global struct to a subclass of
1400         rtl_opt_pass along with...
1401         (pass_data_partition_blocks): ...new pass_data instance and...
1402         (make_pass_partition_blocks): ...new function.
1403         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
1404         struct to a subclass of rtl_opt_pass along with...
1405         (pass_data_branch_target_load_optimize1): ...new pass_data instance
1406         and...
1407         (make_pass_branch_target_load_optimize1): ...new function.
1408         (pass_branch_target_load_optimize2): Convert from a global struct to a
1409         subclass of rtl_opt_pass along with...
1410         (pass_data_branch_target_load_optimize2): ...new pass_data instance
1411         and...
1412         (make_pass_branch_target_load_optimize2): ...new function.
1413         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
1414         of rtl_opt_pass along with...
1415         (pass_data_jump): ...new pass_data instance and...
1416         (make_pass_jump): ...new function.
1417         (pass_jump2): Convert from a global struct to a subclass of
1418         rtl_opt_pass along with...
1419         (pass_data_jump2): ...new pass_data instance and...
1420         (make_pass_jump2): ...new function.
1421         * cfgexpand.c (pass_expand): Convert from a global struct to a
1422         subclass of rtl_opt_pass along with...
1423         (pass_data_expand): ...new pass_data instance and...
1424         (make_pass_expand): ...new function.
1425         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
1426         of rtl_opt_pass along with...
1427         (pass_data_free_cfg): ...new pass_data instance and...
1428         (make_pass_free_cfg): ...new function.
1429         (pass_into_cfg_layout_mode): Convert from a global struct to a
1430         subclass of rtl_opt_pass along with...
1431         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
1432         (make_pass_into_cfg_layout_mode): ...new function.
1433         (pass_outof_cfg_layout_mode): Convert from a global struct to a
1434         subclass of rtl_opt_pass along with...
1435         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
1436         (make_pass_outof_cfg_layout_mode): ...new function.
1437         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
1438         struct to a subclass of gimple_opt_pass along with...
1439         (pass_data_build_cgraph_edges): ...new pass_data instance and...
1440         (make_pass_build_cgraph_edges): ...new function.
1441         (pass_rebuild_cgraph_edges): Convert from a global struct to a
1442         subclass of gimple_opt_pass along with...
1443         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
1444         (make_pass_rebuild_cgraph_edges): ...new function.
1445         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
1446         subclass of gimple_opt_pass along with...
1447         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
1448         and...
1449         (make_pass_remove_cgraph_callee_edges): ...new function.
1450         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
1451         struct to a subclass of rtl_opt_pass along with...
1452         (pass_data_stack_adjustments): ...new pass_data instance and...
1453         (make_pass_stack_adjustments): ...new function.
1454         * combine.c (pass_combine): Convert from a global struct to a subclass
1455         of rtl_opt_pass along with...
1456         (pass_data_combine): ...new pass_data instance and...
1457         (make_pass_combine): ...new function.
1458         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
1459         global struct to a subclass of rtl_opt_pass along with...
1460         (pass_data_compare_elim_after_reload): ...new pass_data instance
1461         and...
1462         (make_pass_compare_elim_after_reload): ...new function.
1463         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
1464         of rtl_opt_pass along with...
1465         (pass_data_rtl_cprop): ...new pass_data instance and...
1466         (make_pass_rtl_cprop): ...new function.
1467         * cse.c (pass_cse): Convert from a global struct to a subclass of
1468         rtl_opt_pass along with...
1469         (pass_data_cse): ...new pass_data instance and...
1470         (make_pass_cse): ...new function.
1471         (pass_cse2): Convert from a global struct to a subclass of
1472         rtl_opt_pass along with...
1473         (pass_data_cse2): ...new pass_data instance and...
1474         (make_pass_cse2): ...new function.
1475         (pass_cse_after_global_opts): Convert from a global struct to a
1476         subclass of rtl_opt_pass along with...
1477         (pass_data_cse_after_global_opts): ...new pass_data instance and...
1478         (make_pass_cse_after_global_opts): ...new function.
1479         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
1480         of rtl_opt_pass along with...
1481         (pass_data_ud_rtl_dce): ...new pass_data instance and...
1482         (make_pass_ud_rtl_dce): ...new function.
1483         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
1484         rtl_opt_pass along with...
1485         (pass_data_fast_rtl_dce): ...new pass_data instance and...
1486         (make_pass_fast_rtl_dce): ...new function.
1487         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
1488         a subclass of rtl_opt_pass along with...
1489         (pass_data_df_initialize_opt): ...new pass_data instance and...
1490         (make_pass_df_initialize_opt): ...new function.
1491         (pass_df_initialize_no_opt): Convert from a global struct to a
1492         subclass of rtl_opt_pass along with...
1493         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
1494         (make_pass_df_initialize_no_opt): ...new function.
1495         (pass_df_finish): Convert from a global struct to a subclass of
1496         rtl_opt_pass along with...
1497         (pass_data_df_finish): ...new pass_data instance and...
1498         (make_pass_df_finish): ...new function.
1499         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
1500         rtl_opt_pass along with...
1501         (pass_data_rtl_dse1): ...new pass_data instance and...
1502         (make_pass_rtl_dse1): ...new function.
1503         (pass_rtl_dse2): Convert from a global struct to a subclass of
1504         rtl_opt_pass along with...
1505         (pass_data_rtl_dse2): ...new pass_data instance and...
1506         (make_pass_rtl_dse2): ...new function.
1507         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
1508         subclass of rtl_opt_pass along with...
1509         (pass_data_dwarf2_frame): ...new pass_data instance and...
1510         (make_pass_dwarf2_frame): ...new function.
1511         * except.c (pass_set_nothrow_function_flags): Convert from a global
1512         struct to a subclass of rtl_opt_pass along with...
1513         (pass_data_set_nothrow_function_flags): ...new pass_data instance
1514         and...
1515         (make_pass_set_nothrow_function_flags): ...new function.
1516         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
1517         subclass of rtl_opt_pass along with...
1518         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
1519         and...
1520         (make_pass_convert_to_eh_region_ranges): ...new function.
1521         * final.c (pass_compute_alignments): Convert from a global struct to a
1522         subclass of rtl_opt_pass along with...
1523         (pass_data_compute_alignments): ...new pass_data instance and...
1524         (make_pass_compute_alignments): ...new function.
1525         (pass_final): Convert from a global struct to a subclass of
1526         rtl_opt_pass along with...
1527         (pass_data_final): ...new pass_data instance and...
1528         (make_pass_final): ...new function.
1529         (pass_shorten_branches): Convert from a global struct to a subclass of
1530         rtl_opt_pass along with...
1531         (pass_data_shorten_branches): ...new pass_data instance and...
1532         (make_pass_shorten_branches): ...new function.
1533         (pass_clean_state): Convert from a global struct to a subclass of
1534         rtl_opt_pass along with...
1535         (pass_data_clean_state): ...new pass_data instance and...
1536         (make_pass_clean_state): ...new function.
1537         * function.c (pass_instantiate_virtual_regs): Convert from a global
1538         struct to a subclass of rtl_opt_pass along with...
1539         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
1540         (make_pass_instantiate_virtual_regs): ...new function.
1541         (pass_leaf_regs): Convert from a global struct to a subclass of
1542         rtl_opt_pass along with...
1543         (pass_data_leaf_regs): ...new pass_data instance and...
1544         (make_pass_leaf_regs): ...new function.
1545         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1546         subclass of rtl_opt_pass along with...
1547         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1548         and...
1549         (make_pass_thread_prologue_and_epilogue): ...new function.
1550         (pass_match_asm_constraints): Convert from a global struct to a
1551         subclass of rtl_opt_pass along with...
1552         (pass_data_match_asm_constraints): ...new pass_data instance and...
1553         (make_pass_match_asm_constraints): ...new function.
1554         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1555         subclass of rtl_opt_pass along with...
1556         (pass_data_rtl_fwprop): ...new pass_data instance and...
1557         (make_pass_rtl_fwprop): ...new function.
1558         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1559         rtl_opt_pass along with...
1560         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1561         (make_pass_rtl_fwprop_addr): ...new function.
1562         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1563         rtl_opt_pass along with...
1564         (pass_data_rtl_pre): ...new pass_data instance and...
1565         (make_pass_rtl_pre): ...new function.
1566         (pass_rtl_hoist): Convert from a global struct to a subclass of
1567         rtl_opt_pass along with...
1568         (pass_data_rtl_hoist): ...new pass_data instance and...
1569         (make_pass_rtl_hoist): ...new function.
1570         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1571         subclass of gimple_opt_pass along with...
1572         (pass_data_lower_cf): ...new pass_data instance and...
1573         (make_pass_lower_cf): ...new function.
1574         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1575         from a global struct to a subclass of gimple_opt_pass along with...
1576         (pass_data_strength_reduction): ...new pass_data instance and...
1577         (make_pass_strength_reduction): ...new function.
1578         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1579         of rtl_opt_pass along with...
1580         (pass_data_rtl_ifcvt): ...new pass_data instance and...
1581         (make_pass_rtl_ifcvt): ...new function.
1582         (pass_if_after_combine): Convert from a global struct to a subclass of
1583         rtl_opt_pass along with...
1584         (pass_data_if_after_combine): ...new pass_data instance and...
1585         (make_pass_if_after_combine): ...new function.
1586         (pass_if_after_reload): Convert from a global struct to a subclass of
1587         rtl_opt_pass along with...
1588         (pass_data_if_after_reload): ...new pass_data instance and...
1589         (make_pass_if_after_reload): ...new function.
1590         * init-regs.c (pass_initialize_regs): Convert from a global struct to
1591         a subclass of rtl_opt_pass along with...
1592         (pass_data_initialize_regs): ...new pass_data instance and...
1593         (make_pass_initialize_regs): ...new function.
1594         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1595         of ipa_opt_pass_d along with...
1596         (pass_data_ipa_cp): ...new pass_data instance and...
1597         (make_pass_ipa_cp): ...new function.
1598         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1599         global struct to a subclass of gimple_opt_pass along with...
1600         (pass_data_inline_parameters): ...new pass_data instance and...
1601         (make_pass_inline_parameters): ...new function.
1602         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1603         subclass of gimple_opt_pass along with...
1604         (pass_data_early_inline): ...new pass_data instance and...
1605         (make_pass_early_inline): ...new function.
1606         (pass_ipa_inline): Convert from a global struct to a subclass of
1607         ipa_opt_pass_d along with...
1608         (pass_data_ipa_inline): ...new pass_data instance and...
1609         (make_pass_ipa_inline): ...new function.
1610         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1611         struct to a subclass of gimple_opt_pass along with...
1612         (pass_data_local_pure_const): ...new pass_data instance and...
1613         (make_pass_local_pure_const): ...new function.
1614         (pass_ipa_pure_const): Convert from a global struct to a subclass of
1615         ipa_opt_pass_d along with...
1616         (pass_data_ipa_pure_const): ...new pass_data instance and...
1617         (make_pass_ipa_pure_const): ...new function.
1618         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1619         to a subclass of ipa_opt_pass_d along with...
1620         (pass_data_ipa_reference): ...new pass_data instance and...
1621         (make_pass_ipa_reference): ...new function.
1622         * ipa-split.c (pass_split_functions): Convert from a global struct to
1623         a subclass of gimple_opt_pass along with...
1624         (pass_data_split_functions): ...new pass_data instance and...
1625         (make_pass_split_functions): ...new function.
1626         (pass_feedback_split_functions): Convert from a global struct to a
1627         subclass of gimple_opt_pass along with...
1628         (pass_data_feedback_split_functions): ...new pass_data instance and...
1629         (make_pass_feedback_split_functions): ...new function.
1630         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1631         global struct to a subclass of simple_ipa_opt_pass along with...
1632         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1633         instance and...
1634         (make_pass_ipa_function_and_variable_visibility): ...new function.
1635         (pass_ipa_free_inline_summary): Convert from a global struct to a
1636         subclass of simple_ipa_opt_pass along with...
1637         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1638         (make_pass_ipa_free_inline_summary): ...new function.
1639         (pass_ipa_whole_program_visibility): Convert from a global struct to a
1640         subclass of ipa_opt_pass_d along with...
1641         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1642         and...
1643         (make_pass_ipa_whole_program_visibility): ...new function.
1644         (pass_ipa_profile): Convert from a global struct to a subclass of
1645         ipa_opt_pass_d along with...
1646         (pass_data_ipa_profile): ...new pass_data instance and...
1647         (make_pass_ipa_profile): ...new function.
1648         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1649         ipa_opt_pass_d along with...
1650         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1651         (make_pass_ipa_cdtor_merge): ...new function.
1652         * ira.c (pass_ira): Convert from a global struct to a subclass of
1653         rtl_opt_pass along with...
1654         (pass_data_ira): ...new pass_data instance and...
1655         (make_pass_ira): ...new function.
1656         (pass_reload): Convert from a global struct to a subclass of
1657         rtl_opt_pass along with...
1658         (pass_data_reload): ...new pass_data instance and...
1659         (make_pass_reload): ...new function.
1660         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1661         subclass of rtl_opt_pass along with...
1662         (pass_data_cleanup_barriers): ...new pass_data instance and...
1663         (make_pass_cleanup_barriers): ...new function.
1664         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1665         of rtl_opt_pass along with...
1666         (pass_data_loop2): ...new pass_data instance and...
1667         (make_pass_loop2): ...new function.
1668         (pass_rtl_loop_init): Convert from a global struct to a subclass of
1669         rtl_opt_pass along with...
1670         (pass_data_rtl_loop_init): ...new pass_data instance and...
1671         (make_pass_rtl_loop_init): ...new function.
1672         (pass_rtl_loop_done): Convert from a global struct to a subclass of
1673         rtl_opt_pass along with...
1674         (pass_data_rtl_loop_done): ...new pass_data instance and...
1675         (make_pass_rtl_loop_done): ...new function.
1676         (pass_rtl_move_loop_invariants): Convert from a global struct to a
1677         subclass of rtl_opt_pass along with...
1678         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1679         (make_pass_rtl_move_loop_invariants): ...new function.
1680         (pass_rtl_unswitch): Convert from a global struct to a subclass of
1681         rtl_opt_pass along with...
1682         (pass_data_rtl_unswitch): ...new pass_data instance and...
1683         (make_pass_rtl_unswitch): ...new function.
1684         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1685         subclass of rtl_opt_pass along with...
1686         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1687         and...
1688         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1689         (pass_rtl_doloop): Convert from a global struct to a subclass of
1690         rtl_opt_pass along with...
1691         (pass_data_rtl_doloop): ...new pass_data instance and...
1692         (make_pass_rtl_doloop): ...new function.
1693         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1694         a subclass of rtl_opt_pass along with...
1695         (pass_data_lower_subreg): ...new pass_data instance and...
1696         (make_pass_lower_subreg): ...new function.
1697         (pass_lower_subreg2): Convert from a global struct to a subclass of
1698         rtl_opt_pass along with...
1699         (pass_data_lower_subreg2): ...new pass_data instance and...
1700         (make_pass_lower_subreg2): ...new function.
1701         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1702         struct to a subclass of ipa_opt_pass_d along with...
1703         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1704         (make_pass_ipa_lto_gimple_out): ...new function.
1705         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1706         of ipa_opt_pass_d along with...
1707         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
1708         (make_pass_ipa_lto_finish_out): ...new function.
1709         * mode-switching.c (pass_mode_switching): Convert from a global struct
1710         to a subclass of rtl_opt_pass along with...
1711         (pass_data_mode_switching): ...new pass_data instance and...
1712         (make_pass_mode_switching): ...new function.
1713         * modulo-sched.c (pass_sms): Convert from a global struct to a
1714         subclass of rtl_opt_pass along with...
1715         (pass_data_sms): ...new pass_data instance and...
1716         (make_pass_sms): ...new function.
1717         * omp-low.c (pass_expand_omp): Convert from a global struct to a
1718         subclass of gimple_opt_pass along with...
1719         (pass_data_expand_omp): ...new pass_data instance and...
1720         (make_pass_expand_omp): ...new function.
1721         (pass_lower_omp): Convert from a global struct to a subclass of
1722         gimple_opt_pass along with...
1723         (pass_data_lower_omp): ...new pass_data instance and...
1724         (make_pass_lower_omp): ...new function.
1725         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
1726         of gimple_opt_pass along with...
1727         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
1728         (make_pass_diagnose_omp_blocks): ...new function.
1729         * passes.c (pass_early_local_passes): Convert from a global struct to
1730         a subclass of simple_ipa_opt_pass along with...
1731         (pass_data_early_local_passes): ...new pass_data instance and...
1732         (make_pass_early_local_passes): ...new function.
1733         (pass_all_early_optimizations): Convert from a global struct to a
1734         subclass of gimple_opt_pass along with...
1735         (pass_data_all_early_optimizations): ...new pass_data instance and...
1736         (make_pass_all_early_optimizations): ...new function.
1737         (pass_all_optimizations): Convert from a global struct to a subclass
1738         of gimple_opt_pass along with...
1739         (pass_data_all_optimizations): ...new pass_data instance and...
1740         (make_pass_all_optimizations): ...new function.
1741         (pass_all_optimizations_g): Convert from a global struct to a subclass
1742         of gimple_opt_pass along with...
1743         (pass_data_all_optimizations_g): ...new pass_data instance and...
1744         (make_pass_all_optimizations_g): ...new function.
1745         (pass_rest_of_compilation): Convert from a global struct to a subclass
1746         of rtl_opt_pass along with...
1747         (pass_data_rest_of_compilation): ...new pass_data instance and...
1748         (make_pass_rest_of_compilation): ...new function.
1749         (pass_postreload): Convert from a global struct to a subclass of
1750         rtl_opt_pass along with...
1751         (pass_data_postreload): ...new pass_data instance and...
1752         (make_pass_postreload): ...new function.
1753         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
1754         subclass of rtl_opt_pass along with...
1755         (pass_data_gcse2): ...new pass_data instance and...
1756         (make_pass_gcse2): ...new function.
1757         * postreload.c (pass_postreload_cse): Convert from a global struct to
1758         a subclass of rtl_opt_pass along with...
1759         (pass_data_postreload_cse): ...new pass_data instance and...
1760         (make_pass_postreload_cse): ...new function.
1761         * predict.c (pass_profile): Convert from a global struct to a subclass
1762         of gimple_opt_pass along with...
1763         (pass_data_profile): ...new pass_data instance and...
1764         (make_pass_profile): ...new function.
1765         (pass_strip_predict_hints): Convert from a global struct to a subclass
1766         of gimple_opt_pass along with...
1767         (pass_data_strip_predict_hints): ...new pass_data instance and...
1768         (make_pass_strip_predict_hints): ...new function.
1769         * recog.c (pass_peephole2): Convert from a global struct to a subclass
1770         of rtl_opt_pass along with...
1771         (pass_data_peephole2): ...new pass_data instance and...
1772         (make_pass_peephole2): ...new function.
1773         (pass_split_all_insns): Convert from a global struct to a subclass of
1774         rtl_opt_pass along with...
1775         (pass_data_split_all_insns): ...new pass_data instance and...
1776         (make_pass_split_all_insns): ...new function.
1777         (pass_split_after_reload): Convert from a global struct to a subclass
1778         of rtl_opt_pass along with...
1779         (pass_data_split_after_reload): ...new pass_data instance and...
1780         (make_pass_split_after_reload): ...new function.
1781         (pass_split_before_regstack): Convert from a global struct to a
1782         subclass of rtl_opt_pass along with...
1783         (pass_data_split_before_regstack): ...new pass_data instance and...
1784         (make_pass_split_before_regstack): ...new function.
1785         (pass_split_before_sched2): Convert from a global struct to a subclass
1786         of rtl_opt_pass along with...
1787         (pass_data_split_before_sched2): ...new pass_data instance and...
1788         (make_pass_split_before_sched2): ...new function.
1789         (pass_split_for_shorten_branches): Convert from a global struct to a
1790         subclass of rtl_opt_pass along with...
1791         (pass_data_split_for_shorten_branches): ...new pass_data instance
1792         and...
1793         (make_pass_split_for_shorten_branches): ...new function.
1794         * ree.c (pass_ree): Convert from a global struct to a subclass of
1795         rtl_opt_pass along with...
1796         (pass_data_ree): ...new pass_data instance and...
1797         (make_pass_ree): ...new function.
1798         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
1799         subclass of rtl_opt_pass along with...
1800         (pass_data_stack_regs): ...new pass_data instance and...
1801         (make_pass_stack_regs): ...new function.
1802         (pass_stack_regs_run): Convert from a global struct to a subclass of
1803         rtl_opt_pass along with...
1804         (pass_data_stack_regs_run): ...new pass_data instance and...
1805         (make_pass_stack_regs_run): ...new function.
1806         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1807         subclass of rtl_opt_pass along with...
1808         (pass_data_cprop_hardreg): ...new pass_data instance and...
1809         (make_pass_cprop_hardreg): ...new function.
1810         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1811         subclass of rtl_opt_pass along with...
1812         (pass_data_reginfo_init): ...new pass_data instance and...
1813         (make_pass_reginfo_init): ...new function.
1814         * regmove.c (pass_regmove): Convert from a global struct to a subclass
1815         of rtl_opt_pass along with...
1816         (pass_data_regmove): ...new pass_data instance and...
1817         (make_pass_regmove): ...new function.
1818         * regrename.c (pass_regrename): Convert from a global struct to a
1819         subclass of rtl_opt_pass along with...
1820         (pass_data_regrename): ...new pass_data instance and...
1821         (make_pass_regrename): ...new function.
1822         * reorg.c (pass_delay_slots): Convert from a global struct to a
1823         subclass of rtl_opt_pass along with...
1824         (pass_data_delay_slots): ...new pass_data instance and...
1825         (make_pass_delay_slots): ...new function.
1826         (pass_machine_reorg): Convert from a global struct to a subclass of
1827         rtl_opt_pass along with...
1828         (pass_data_machine_reorg): ...new pass_data instance and...
1829         (make_pass_machine_reorg): ...new function.
1830         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1831         of rtl_opt_pass along with...
1832         (pass_data_sched): ...new pass_data instance and...
1833         (make_pass_sched): ...new function.
1834         (pass_sched2): Convert from a global struct to a subclass of
1835         rtl_opt_pass along with...
1836         (pass_data_sched2): ...new pass_data instance and...
1837         (make_pass_sched2): ...new function.
1838         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1839         to a subclass of rtl_opt_pass along with...
1840         (pass_data_stack_ptr_mod): ...new pass_data instance and...
1841         (make_pass_stack_ptr_mod): ...new function.
1842         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1843         to a subclass of rtl_opt_pass along with...
1844         (pass_data_rtl_store_motion): ...new pass_data instance and...
1845         (make_pass_rtl_store_motion): ...new function.
1846         * tracer.c (pass_tracer): Convert from a global struct to a subclass
1847         of gimple_opt_pass along with...
1848         (pass_data_tracer): ...new pass_data instance and...
1849         (make_pass_tracer): ...new function.
1850         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1851         to a subclass of gimple_opt_pass along with...
1852         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1853         (make_pass_diagnose_tm_blocks): ...new function.
1854         (pass_lower_tm): Convert from a global struct to a subclass of
1855         gimple_opt_pass along with...
1856         (pass_data_lower_tm): ...new pass_data instance and...
1857         (make_pass_lower_tm): ...new function.
1858         (pass_tm_init): Convert from a global struct to a subclass of
1859         gimple_opt_pass along with...
1860         (pass_data_tm_init): ...new pass_data instance and...
1861         (make_pass_tm_init): ...new function.
1862         (pass_tm_mark): Convert from a global struct to a subclass of
1863         gimple_opt_pass along with...
1864         (pass_data_tm_mark): ...new pass_data instance and...
1865         (make_pass_tm_mark): ...new function.
1866         (pass_tm_edges): Convert from a global struct to a subclass of
1867         gimple_opt_pass along with...
1868         (pass_data_tm_edges): ...new pass_data instance and...
1869         (make_pass_tm_edges): ...new function.
1870         (pass_tm_memopt): Convert from a global struct to a subclass of
1871         gimple_opt_pass along with...
1872         (pass_data_tm_memopt): ...new pass_data instance and...
1873         (make_pass_tm_memopt): ...new function.
1874         (pass_ipa_tm): Convert from a global struct to a subclass of
1875         simple_ipa_opt_pass along with...
1876         (pass_data_ipa_tm): ...new pass_data instance and...
1877         (make_pass_ipa_tm): ...new function.
1878         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1879         subclass of gimple_opt_pass along with...
1880         (pass_data_call_cdce): ...new pass_data instance and...
1881         (make_pass_call_cdce): ...new function.
1882         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1883         subclass of gimple_opt_pass along with...
1884         (pass_data_build_cfg): ...new pass_data instance and...
1885         (make_pass_build_cfg): ...new function.
1886         (pass_split_crit_edges): Convert from a global struct to a subclass of
1887         gimple_opt_pass along with...
1888         (pass_data_split_crit_edges): ...new pass_data instance and...
1889         (make_pass_split_crit_edges): ...new function.
1890         (pass_warn_function_return): Convert from a global struct to a
1891         subclass of gimple_opt_pass along with...
1892         (pass_data_warn_function_return): ...new pass_data instance and...
1893         (make_pass_warn_function_return): ...new function.
1894         (pass_warn_function_noreturn): Convert from a global struct to a
1895         subclass of gimple_opt_pass along with...
1896         (pass_data_warn_function_noreturn): ...new pass_data instance and...
1897         (make_pass_warn_function_noreturn): ...new function.
1898         (pass_warn_unused_result): Convert from a global struct to a subclass
1899         of gimple_opt_pass along with...
1900         (pass_data_warn_unused_result): ...new pass_data instance and...
1901         (make_pass_warn_unused_result): ...new function.
1902         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1903         a subclass of gimple_opt_pass along with...
1904         (pass_data_merge_phi): ...new pass_data instance and...
1905         (make_pass_merge_phi): ...new function.
1906         * tree-complex.c (pass_lower_complex): Convert from a global struct to
1907         a subclass of gimple_opt_pass along with...
1908         (pass_data_lower_complex): ...new pass_data instance and...
1909         (make_pass_lower_complex): ...new function.
1910         (pass_lower_complex_O0): Convert from a global struct to a subclass of
1911         gimple_opt_pass along with...
1912         (pass_data_lower_complex_O0): ...new pass_data instance and...
1913         (make_pass_lower_complex_O0): ...new function.
1914         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1915         subclass of gimple_opt_pass along with...
1916         (pass_data_lower_eh): ...new pass_data instance and...
1917         (make_pass_lower_eh): ...new function.
1918         (pass_refactor_eh): Convert from a global struct to a subclass of
1919         gimple_opt_pass along with...
1920         (pass_data_refactor_eh): ...new pass_data instance and...
1921         (make_pass_refactor_eh): ...new function.
1922         (pass_lower_resx): Convert from a global struct to a subclass of
1923         gimple_opt_pass along with...
1924         (pass_data_lower_resx): ...new pass_data instance and...
1925         (make_pass_lower_resx): ...new function.
1926         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1927         of gimple_opt_pass along with...
1928         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1929         (make_pass_lower_eh_dispatch): ...new function.
1930         (pass_cleanup_eh): Convert from a global struct to a subclass of
1931         gimple_opt_pass along with...
1932         (pass_data_cleanup_eh): ...new pass_data instance and...
1933         (make_pass_cleanup_eh): ...new function.
1934         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1935         to a subclass of simple_ipa_opt_pass along with...
1936         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1937         (make_pass_ipa_lower_emutls): ...new function.
1938         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1939         a subclass of gimple_opt_pass along with...
1940         (pass_data_if_conversion): ...new pass_data instance and...
1941         (make_pass_if_conversion): ...new function.
1942         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1943         subclass of gimple_opt_pass along with...
1944         (pass_data_build_ssa): ...new pass_data instance and...
1945         (make_pass_build_ssa): ...new function.
1946         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1947         global struct to a subclass of gimple_opt_pass along with...
1948         (pass_data_loop_distribution): ...new pass_data instance and...
1949         (make_pass_loop_distribution): ...new function.
1950         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1951         subclass of gimple_opt_pass along with...
1952         (pass_data_mudflap_1): ...new pass_data instance and...
1953         (make_pass_mudflap_1): ...new function.
1954         (pass_mudflap_2): Convert from a global struct to a subclass of
1955         gimple_opt_pass along with...
1956         (pass_data_mudflap_2): ...new pass_data instance and...
1957         (make_pass_mudflap_2): ...new function.
1958         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1959         subclass of gimple_opt_pass along with...
1960         (pass_data_mudflap_1): ...new pass_data instance and...
1961         (make_pass_mudflap_1): ...new function.
1962         (pass_mudflap_2): Convert from a global struct to a subclass of
1963         gimple_opt_pass along with...
1964         (pass_data_mudflap_2): ...new pass_data instance and...
1965         (make_pass_mudflap_2): ...new function.
1966         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1967         gimple_opt_pass along with...
1968         (pass_data_nrv): ...new pass_data instance and...
1969         (make_pass_nrv): ...new function.
1970         (pass_return_slot): Convert from a global struct to a subclass of
1971         gimple_opt_pass along with...
1972         (pass_data_return_slot): ...new pass_data instance and...
1973         (make_pass_return_slot): ...new function.
1974         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1975         to a subclass of gimple_opt_pass along with...
1976         (pass_data_object_sizes): ...new pass_data instance and...
1977         (make_pass_object_sizes): ...new function.
1978         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1979         global struct to a subclass of gimple_opt_pass along with...
1980         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1981         and...
1982         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1983         (pass_fixup_cfg): Convert from a global struct to a subclass of
1984         gimple_opt_pass along with...
1985         (pass_data_fixup_cfg): ...new pass_data instance and...
1986         (make_pass_fixup_cfg): ...new function.
1987         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1988         (make_pass_mudflap_1): ...new function.
1989         (pass_mudflap_2): Replace declaration with that of...
1990         (make_pass_mudflap_2): ...new function.
1991         (pass_asan): Replace declaration with that of...
1992         (make_pass_asan): ...new function.
1993         (pass_asan_O0): Replace declaration with that of...
1994         (make_pass_asan_O0): ...new function.
1995         (pass_tsan): Replace declaration with that of...
1996         (make_pass_tsan): ...new function.
1997         (pass_tsan_O0): Replace declaration with that of...
1998         (make_pass_tsan_O0): ...new function.
1999         (pass_lower_cf): Replace declaration with that of...
2000         (make_pass_lower_cf): ...new function.
2001         (pass_refactor_eh): Replace declaration with that of...
2002         (make_pass_refactor_eh): ...new function.
2003         (pass_lower_eh): Replace declaration with that of...
2004         (make_pass_lower_eh): ...new function.
2005         (pass_lower_eh_dispatch): Replace declaration with that of...
2006         (make_pass_lower_eh_dispatch): ...new function.
2007         (pass_lower_resx): Replace declaration with that of...
2008         (make_pass_lower_resx): ...new function.
2009         (pass_build_cfg): Replace declaration with that of...
2010         (make_pass_build_cfg): ...new function.
2011         (pass_early_tree_profile): Replace declaration with that of...
2012         (make_pass_early_tree_profile): ...new function.
2013         (pass_cleanup_eh): Replace declaration with that of...
2014         (make_pass_cleanup_eh): ...new function.
2015         (pass_sra): Replace declaration with that of...
2016         (make_pass_sra): ...new function.
2017         (pass_sra_early): Replace declaration with that of...
2018         (make_pass_sra_early): ...new function.
2019         (pass_early_ipa_sra): Replace declaration with that of...
2020         (make_pass_early_ipa_sra): ...new function.
2021         (pass_tail_recursion): Replace declaration with that of...
2022         (make_pass_tail_recursion): ...new function.
2023         (pass_tail_calls): Replace declaration with that of...
2024         (make_pass_tail_calls): ...new function.
2025         (pass_tree_loop): Replace declaration with that of...
2026         (make_pass_tree_loop): ...new function.
2027         (pass_tree_loop_init): Replace declaration with that of...
2028         (make_pass_tree_loop_init): ...new function.
2029         (pass_lim): Replace declaration with that of...
2030         (make_pass_lim): ...new function.
2031         (pass_tree_unswitch): Replace declaration with that of...
2032         (make_pass_tree_unswitch): ...new function.
2033         (pass_predcom): Replace declaration with that of...
2034         (make_pass_predcom): ...new function.
2035         (pass_iv_canon): Replace declaration with that of...
2036         (make_pass_iv_canon): ...new function.
2037         (pass_scev_cprop): Replace declaration with that of...
2038         (make_pass_scev_cprop): ...new function.
2039         (pass_empty_loop): Replace declaration with that of...
2040         (make_pass_empty_loop): ...new function.
2041         (pass_record_bounds): Replace declaration with that of...
2042         (make_pass_record_bounds): ...new function.
2043         (pass_graphite): Replace declaration with that of...
2044         (make_pass_graphite): ...new function.
2045         (pass_graphite_transforms): Replace declaration with that of...
2046         (make_pass_graphite_transforms): ...new function.
2047         (pass_if_conversion): Replace declaration with that of...
2048         (make_pass_if_conversion): ...new function.
2049         (pass_loop_distribution): Replace declaration with that of...
2050         (make_pass_loop_distribution): ...new function.
2051         (pass_vectorize): Replace declaration with that of...
2052         (make_pass_vectorize): ...new function.
2053         (pass_slp_vectorize): Replace declaration with that of...
2054         (make_pass_slp_vectorize): ...new function.
2055         (pass_complete_unroll): Replace declaration with that of...
2056         (make_pass_complete_unroll): ...new function.
2057         (pass_complete_unrolli): Replace declaration with that of...
2058         (make_pass_complete_unrolli): ...new function.
2059         (pass_parallelize_loops): Replace declaration with that of...
2060         (make_pass_parallelize_loops): ...new function.
2061         (pass_loop_prefetch): Replace declaration with that of...
2062         (make_pass_loop_prefetch): ...new function.
2063         (pass_iv_optimize): Replace declaration with that of...
2064         (make_pass_iv_optimize): ...new function.
2065         (pass_tree_loop_done): Replace declaration with that of...
2066         (make_pass_tree_loop_done): ...new function.
2067         (pass_ch): Replace declaration with that of...
2068         (make_pass_ch): ...new function.
2069         (pass_ccp): Replace declaration with that of...
2070         (make_pass_ccp): ...new function.
2071         (pass_phi_only_cprop): Replace declaration with that of...
2072         (make_pass_phi_only_cprop): ...new function.
2073         (pass_build_ssa): Replace declaration with that of...
2074         (make_pass_build_ssa): ...new function.
2075         (pass_build_alias): Replace declaration with that of...
2076         (make_pass_build_alias): ...new function.
2077         (pass_build_ealias): Replace declaration with that of...
2078         (make_pass_build_ealias): ...new function.
2079         (pass_dominator): Replace declaration with that of...
2080         (make_pass_dominator): ...new function.
2081         (pass_dce): Replace declaration with that of...
2082         (make_pass_dce): ...new function.
2083         (pass_dce_loop): Replace declaration with that of...
2084         (make_pass_dce_loop): ...new function.
2085         (pass_cd_dce): Replace declaration with that of...
2086         (make_pass_cd_dce): ...new function.
2087         (pass_call_cdce): Replace declaration with that of...
2088         (make_pass_call_cdce): ...new function.
2089         (pass_merge_phi): Replace declaration with that of...
2090         (make_pass_merge_phi): ...new function.
2091         (pass_split_crit_edges): Replace declaration with that of...
2092         (make_pass_split_crit_edges): ...new function.
2093         (pass_pre): Replace declaration with that of...
2094         (make_pass_pre): ...new function.
2095         (pass_profile): Replace declaration with that of...
2096         (make_pass_profile): ...new function.
2097         (pass_strip_predict_hints): Replace declaration with that of...
2098         (make_pass_strip_predict_hints): ...new function.
2099         (pass_lower_complex_O0): Replace declaration with that of...
2100         (make_pass_lower_complex_O0): ...new function.
2101         (pass_lower_complex): Replace declaration with that of...
2102         (make_pass_lower_complex): ...new function.
2103         (pass_lower_vector): Replace declaration with that of...
2104         (make_pass_lower_vector): ...new function.
2105         (pass_lower_vector_ssa): Replace declaration with that of...
2106         (make_pass_lower_vector_ssa): ...new function.
2107         (pass_lower_omp): Replace declaration with that of...
2108         (make_pass_lower_omp): ...new function.
2109         (pass_diagnose_omp_blocks): Replace declaration with that of...
2110         (make_pass_diagnose_omp_blocks): ...new function.
2111         (pass_expand_omp): Replace declaration with that of...
2112         (make_pass_expand_omp): ...new function.
2113         (pass_expand_omp_ssa): Replace declaration with that of...
2114         (make_pass_expand_omp_ssa): ...new function.
2115         (pass_object_sizes): Replace declaration with that of...
2116         (make_pass_object_sizes): ...new function.
2117         (pass_strlen): Replace declaration with that of...
2118         (make_pass_strlen): ...new function.
2119         (pass_fold_builtins): Replace declaration with that of...
2120         (make_pass_fold_builtins): ...new function.
2121         (pass_stdarg): Replace declaration with that of...
2122         (make_pass_stdarg): ...new function.
2123         (pass_early_warn_uninitialized): Replace declaration with that of...
2124         (make_pass_early_warn_uninitialized): ...new function.
2125         (pass_late_warn_uninitialized): Replace declaration with that of...
2126         (make_pass_late_warn_uninitialized): ...new function.
2127         (pass_cse_reciprocals): Replace declaration with that of...
2128         (make_pass_cse_reciprocals): ...new function.
2129         (pass_cse_sincos): Replace declaration with that of...
2130         (make_pass_cse_sincos): ...new function.
2131         (pass_optimize_bswap): Replace declaration with that of...
2132         (make_pass_optimize_bswap): ...new function.
2133         (pass_optimize_widening_mul): Replace declaration with that of...
2134         (make_pass_optimize_widening_mul): ...new function.
2135         (pass_warn_function_return): Replace declaration with that of...
2136         (make_pass_warn_function_return): ...new function.
2137         (pass_warn_function_noreturn): Replace declaration with that of...
2138         (make_pass_warn_function_noreturn): ...new function.
2139         (pass_cselim): Replace declaration with that of...
2140         (make_pass_cselim): ...new function.
2141         (pass_phiopt): Replace declaration with that of...
2142         (make_pass_phiopt): ...new function.
2143         (pass_forwprop): Replace declaration with that of...
2144         (make_pass_forwprop): ...new function.
2145         (pass_phiprop): Replace declaration with that of...
2146         (make_pass_phiprop): ...new function.
2147         (pass_tree_ifcombine): Replace declaration with that of...
2148         (make_pass_tree_ifcombine): ...new function.
2149         (pass_dse): Replace declaration with that of...
2150         (make_pass_dse): ...new function.
2151         (pass_nrv): Replace declaration with that of...
2152         (make_pass_nrv): ...new function.
2153         (pass_rename_ssa_copies): Replace declaration with that of...
2154         (make_pass_rename_ssa_copies): ...new function.
2155         (pass_sink_code): Replace declaration with that of...
2156         (make_pass_sink_code): ...new function.
2157         (pass_fre): Replace declaration with that of...
2158         (make_pass_fre): ...new function.
2159         (pass_check_data_deps): Replace declaration with that of...
2160         (make_pass_check_data_deps): ...new function.
2161         (pass_copy_prop): Replace declaration with that of...
2162         (make_pass_copy_prop): ...new function.
2163         (pass_vrp): Replace declaration with that of...
2164         (make_pass_vrp): ...new function.
2165         (pass_uncprop): Replace declaration with that of...
2166         (make_pass_uncprop): ...new function.
2167         (pass_return_slot): Replace declaration with that of...
2168         (make_pass_return_slot): ...new function.
2169         (pass_reassoc): Replace declaration with that of...
2170         (make_pass_reassoc): ...new function.
2171         (pass_rebuild_cgraph_edges): Replace declaration with that of...
2172         (make_pass_rebuild_cgraph_edges): ...new function.
2173         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2174         (make_pass_remove_cgraph_callee_edges): ...new function.
2175         (pass_build_cgraph_edges): Replace declaration with that of...
2176         (make_pass_build_cgraph_edges): ...new function.
2177         (pass_local_pure_const): Replace declaration with that of...
2178         (make_pass_local_pure_const): ...new function.
2179         (pass_tracer): Replace declaration with that of...
2180         (make_pass_tracer): ...new function.
2181         (pass_warn_unused_result): Replace declaration with that of...
2182         (make_pass_warn_unused_result): ...new function.
2183         (pass_diagnose_tm_blocks): Replace declaration with that of...
2184         (make_pass_diagnose_tm_blocks): ...new function.
2185         (pass_lower_tm): Replace declaration with that of...
2186         (make_pass_lower_tm): ...new function.
2187         (pass_tm_init): Replace declaration with that of...
2188         (make_pass_tm_init): ...new function.
2189         (pass_tm_mark): Replace declaration with that of...
2190         (make_pass_tm_mark): ...new function.
2191         (pass_tm_memopt): Replace declaration with that of...
2192         (make_pass_tm_memopt): ...new function.
2193         (pass_tm_edges): Replace declaration with that of...
2194         (make_pass_tm_edges): ...new function.
2195         (pass_split_functions): Replace declaration with that of...
2196         (make_pass_split_functions): ...new function.
2197         (pass_feedback_split_functions): Replace declaration with that of...
2198         (make_pass_feedback_split_functions): ...new function.
2199         (pass_strength_reduction): Replace declaration with that of...
2200         (make_pass_strength_reduction): ...new function.
2201         (pass_ipa_lower_emutls): Replace declaration with that of...
2202         (make_pass_ipa_lower_emutls): ...new function.
2203         (pass_ipa_function_and_variable_visibility): Replace declaration with
2204         that of...
2205         (make_pass_ipa_function_and_variable_visibility): ...new function.
2206         (pass_ipa_tree_profile): Replace declaration with that of...
2207         (make_pass_ipa_tree_profile): ...new function.
2208         (pass_early_local_passes): Replace declaration with that of...
2209         (make_pass_early_local_passes): ...new function.
2210         (pass_ipa_whole_program_visibility): Replace declaration with that
2211         of...
2212         (make_pass_ipa_whole_program_visibility): ...new function.
2213         (pass_ipa_lto_gimple_out): Replace declaration with that of...
2214         (make_pass_ipa_lto_gimple_out): ...new function.
2215         (pass_ipa_increase_alignment): Replace declaration with that of...
2216         (make_pass_ipa_increase_alignment): ...new function.
2217         (pass_ipa_inline): Replace declaration with that of...
2218         (make_pass_ipa_inline): ...new function.
2219         (pass_ipa_free_lang_data): Replace declaration with that of...
2220         (make_pass_ipa_free_lang_data): ...new function.
2221         (pass_ipa_free_inline_summary): Replace declaration with that of...
2222         (make_pass_ipa_free_inline_summary): ...new function.
2223         (pass_ipa_cp): Replace declaration with that of...
2224         (make_pass_ipa_cp): ...new function.
2225         (pass_ipa_reference): Replace declaration with that of...
2226         (make_pass_ipa_reference): ...new function.
2227         (pass_ipa_pure_const): Replace declaration with that of...
2228         (make_pass_ipa_pure_const): ...new function.
2229         (pass_ipa_pta): Replace declaration with that of...
2230         (make_pass_ipa_pta): ...new function.
2231         (pass_ipa_lto_finish_out): Replace declaration with that of...
2232         (make_pass_ipa_lto_finish_out): ...new function.
2233         (pass_ipa_tm): Replace declaration with that of...
2234         (make_pass_ipa_tm): ...new function.
2235         (pass_ipa_profile): Replace declaration with that of...
2236         (make_pass_ipa_profile): ...new function.
2237         (pass_ipa_cdtor_merge): Replace declaration with that of...
2238         (make_pass_ipa_cdtor_merge): ...new function.
2239         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2240         of...
2241         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2242         (pass_init_datastructures): Replace declaration with that of...
2243         (make_pass_init_datastructures): ...new function.
2244         (pass_fixup_cfg): Replace declaration with that of...
2245         (make_pass_fixup_cfg): ...new function.
2246         (pass_expand): Replace declaration with that of...
2247         (make_pass_expand): ...new function.
2248         (pass_instantiate_virtual_regs): Replace declaration with that of...
2249         (make_pass_instantiate_virtual_regs): ...new function.
2250         (pass_rtl_fwprop): Replace declaration with that of...
2251         (make_pass_rtl_fwprop): ...new function.
2252         (pass_rtl_fwprop_addr): Replace declaration with that of...
2253         (make_pass_rtl_fwprop_addr): ...new function.
2254         (pass_jump): Replace declaration with that of...
2255         (make_pass_jump): ...new function.
2256         (pass_jump2): Replace declaration with that of...
2257         (make_pass_jump2): ...new function.
2258         (pass_lower_subreg): Replace declaration with that of...
2259         (make_pass_lower_subreg): ...new function.
2260         (pass_cse): Replace declaration with that of...
2261         (make_pass_cse): ...new function.
2262         (pass_fast_rtl_dce): Replace declaration with that of...
2263         (make_pass_fast_rtl_dce): ...new function.
2264         (pass_ud_rtl_dce): Replace declaration with that of...
2265         (make_pass_ud_rtl_dce): ...new function.
2266         (pass_rtl_dce): Replace declaration with that of...
2267         (make_pass_rtl_dce): ...new function.
2268         (pass_rtl_dse1): Replace declaration with that of...
2269         (make_pass_rtl_dse1): ...new function.
2270         (pass_rtl_dse2): Replace declaration with that of...
2271         (make_pass_rtl_dse2): ...new function.
2272         (pass_rtl_dse3): Replace declaration with that of...
2273         (make_pass_rtl_dse3): ...new function.
2274         (pass_rtl_cprop): Replace declaration with that of...
2275         (make_pass_rtl_cprop): ...new function.
2276         (pass_rtl_pre): Replace declaration with that of...
2277         (make_pass_rtl_pre): ...new function.
2278         (pass_rtl_hoist): Replace declaration with that of...
2279         (make_pass_rtl_hoist): ...new function.
2280         (pass_rtl_store_motion): Replace declaration with that of...
2281         (make_pass_rtl_store_motion): ...new function.
2282         (pass_cse_after_global_opts): Replace declaration with that of...
2283         (make_pass_cse_after_global_opts): ...new function.
2284         (pass_rtl_ifcvt): Replace declaration with that of...
2285         (make_pass_rtl_ifcvt): ...new function.
2286         (pass_into_cfg_layout_mode): Replace declaration with that of...
2287         (make_pass_into_cfg_layout_mode): ...new function.
2288         (pass_outof_cfg_layout_mode): Replace declaration with that of...
2289         (make_pass_outof_cfg_layout_mode): ...new function.
2290         (pass_loop2): Replace declaration with that of...
2291         (make_pass_loop2): ...new function.
2292         (pass_rtl_loop_init): Replace declaration with that of...
2293         (make_pass_rtl_loop_init): ...new function.
2294         (pass_rtl_move_loop_invariants): Replace declaration with that of...
2295         (make_pass_rtl_move_loop_invariants): ...new function.
2296         (pass_rtl_unswitch): Replace declaration with that of...
2297         (make_pass_rtl_unswitch): ...new function.
2298         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2299         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2300         (pass_rtl_doloop): Replace declaration with that of...
2301         (make_pass_rtl_doloop): ...new function.
2302         (pass_rtl_loop_done): Replace declaration with that of...
2303         (make_pass_rtl_loop_done): ...new function.
2304         (pass_web): Replace declaration with that of...
2305         (make_pass_web): ...new function.
2306         (pass_cse2): Replace declaration with that of...
2307         (make_pass_cse2): ...new function.
2308         (pass_df_initialize_opt): Replace declaration with that of...
2309         (make_pass_df_initialize_opt): ...new function.
2310         (pass_df_initialize_no_opt): Replace declaration with that of...
2311         (make_pass_df_initialize_no_opt): ...new function.
2312         (pass_reginfo_init): Replace declaration with that of...
2313         (make_pass_reginfo_init): ...new function.
2314         (pass_inc_dec): Replace declaration with that of...
2315         (make_pass_inc_dec): ...new function.
2316         (pass_stack_ptr_mod): Replace declaration with that of...
2317         (make_pass_stack_ptr_mod): ...new function.
2318         (pass_initialize_regs): Replace declaration with that of...
2319         (make_pass_initialize_regs): ...new function.
2320         (pass_combine): Replace declaration with that of...
2321         (make_pass_combine): ...new function.
2322         (pass_if_after_combine): Replace declaration with that of...
2323         (make_pass_if_after_combine): ...new function.
2324         (pass_ree): Replace declaration with that of...
2325         (make_pass_ree): ...new function.
2326         (pass_partition_blocks): Replace declaration with that of...
2327         (make_pass_partition_blocks): ...new function.
2328         (pass_match_asm_constraints): Replace declaration with that of...
2329         (make_pass_match_asm_constraints): ...new function.
2330         (pass_regmove): Replace declaration with that of...
2331         (make_pass_regmove): ...new function.
2332         (pass_split_all_insns): Replace declaration with that of...
2333         (make_pass_split_all_insns): ...new function.
2334         (pass_fast_rtl_byte_dce): Replace declaration with that of...
2335         (make_pass_fast_rtl_byte_dce): ...new function.
2336         (pass_lower_subreg2): Replace declaration with that of...
2337         (make_pass_lower_subreg2): ...new function.
2338         (pass_mode_switching): Replace declaration with that of...
2339         (make_pass_mode_switching): ...new function.
2340         (pass_sms): Replace declaration with that of...
2341         (make_pass_sms): ...new function.
2342         (pass_sched): Replace declaration with that of...
2343         (make_pass_sched): ...new function.
2344         (pass_ira): Replace declaration with that of...
2345         (make_pass_ira): ...new function.
2346         (pass_reload): Replace declaration with that of...
2347         (make_pass_reload): ...new function.
2348         (pass_clean_state): Replace declaration with that of...
2349         (make_pass_clean_state): ...new function.
2350         (pass_branch_prob): Replace declaration with that of...
2351         (make_pass_branch_prob): ...new function.
2352         (pass_value_profile_transformations): Replace declaration with that
2353         of...
2354         (make_pass_value_profile_transformations): ...new function.
2355         (pass_postreload_cse): Replace declaration with that of...
2356         (make_pass_postreload_cse): ...new function.
2357         (pass_gcse2): Replace declaration with that of...
2358         (make_pass_gcse2): ...new function.
2359         (pass_split_after_reload): Replace declaration with that of...
2360         (make_pass_split_after_reload): ...new function.
2361         (pass_branch_target_load_optimize1): Replace declaration with that
2362         of...
2363         (make_pass_branch_target_load_optimize1): ...new function.
2364         (pass_thread_prologue_and_epilogue): Replace declaration with that
2365         of...
2366         (make_pass_thread_prologue_and_epilogue): ...new function.
2367         (pass_stack_adjustments): Replace declaration with that of...
2368         (make_pass_stack_adjustments): ...new function.
2369         (pass_peephole2): Replace declaration with that of...
2370         (make_pass_peephole2): ...new function.
2371         (pass_if_after_reload): Replace declaration with that of...
2372         (make_pass_if_after_reload): ...new function.
2373         (pass_regrename): Replace declaration with that of...
2374         (make_pass_regrename): ...new function.
2375         (pass_cprop_hardreg): Replace declaration with that of...
2376         (make_pass_cprop_hardreg): ...new function.
2377         (pass_reorder_blocks): Replace declaration with that of...
2378         (make_pass_reorder_blocks): ...new function.
2379         (pass_branch_target_load_optimize2): Replace declaration with that
2380         of...
2381         (make_pass_branch_target_load_optimize2): ...new function.
2382         (pass_leaf_regs): Replace declaration with that of...
2383         (make_pass_leaf_regs): ...new function.
2384         (pass_split_before_sched2): Replace declaration with that of...
2385         (make_pass_split_before_sched2): ...new function.
2386         (pass_compare_elim_after_reload): Replace declaration with that of...
2387         (make_pass_compare_elim_after_reload): ...new function.
2388         (pass_sched2): Replace declaration with that of...
2389         (make_pass_sched2): ...new function.
2390         (pass_stack_regs): Replace declaration with that of...
2391         (make_pass_stack_regs): ...new function.
2392         (pass_stack_regs_run): Replace declaration with that of...
2393         (make_pass_stack_regs_run): ...new function.
2394         (pass_df_finish): Replace declaration with that of...
2395         (make_pass_df_finish): ...new function.
2396         (pass_compute_alignments): Replace declaration with that of...
2397         (make_pass_compute_alignments): ...new function.
2398         (pass_duplicate_computed_gotos): Replace declaration with that of...
2399         (make_pass_duplicate_computed_gotos): ...new function.
2400         (pass_variable_tracking): Replace declaration with that of...
2401         (make_pass_variable_tracking): ...new function.
2402         (pass_free_cfg): Replace declaration with that of...
2403         (make_pass_free_cfg): ...new function.
2404         (pass_machine_reorg): Replace declaration with that of...
2405         (make_pass_machine_reorg): ...new function.
2406         (pass_cleanup_barriers): Replace declaration with that of...
2407         (make_pass_cleanup_barriers): ...new function.
2408         (pass_delay_slots): Replace declaration with that of...
2409         (make_pass_delay_slots): ...new function.
2410         (pass_split_for_shorten_branches): Replace declaration with that of...
2411         (make_pass_split_for_shorten_branches): ...new function.
2412         (pass_split_before_regstack): Replace declaration with that of...
2413         (make_pass_split_before_regstack): ...new function.
2414         (pass_convert_to_eh_region_ranges): Replace declaration with that
2415         of...
2416         (make_pass_convert_to_eh_region_ranges): ...new function.
2417         (pass_shorten_branches): Replace declaration with that of...
2418         (make_pass_shorten_branches): ...new function.
2419         (pass_set_nothrow_function_flags): Replace declaration with that of...
2420         (make_pass_set_nothrow_function_flags): ...new function.
2421         (pass_dwarf2_frame): Replace declaration with that of...
2422         (make_pass_dwarf2_frame): ...new function.
2423         (pass_final): Replace declaration with that of...
2424         (make_pass_final): ...new function.
2425         (pass_rtl_seqabstr): Replace declaration with that of...
2426         (make_pass_rtl_seqabstr): ...new function.
2427         (pass_release_ssa_names): Replace declaration with that of...
2428         (make_pass_release_ssa_names): ...new function.
2429         (pass_early_inline): Replace declaration with that of...
2430         (make_pass_early_inline): ...new function.
2431         (pass_inline_parameters): Replace declaration with that of...
2432         (make_pass_inline_parameters): ...new function.
2433         (pass_update_address_taken): Replace declaration with that of...
2434         (make_pass_update_address_taken): ...new function.
2435         (pass_convert_switch): Replace declaration with that of...
2436         (make_pass_convert_switch): ...new function.
2437         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
2438         to a subclass of simple_ipa_opt_pass along with...
2439         (pass_data_ipa_tree_profile): ...new pass_data instance and...
2440         (make_pass_ipa_tree_profile): ...new function.
2441         * tree-sra.c (pass_sra_early): Convert from a global struct to a
2442         subclass of gimple_opt_pass along with...
2443         (pass_data_sra_early): ...new pass_data instance and...
2444         (make_pass_sra_early): ...new function.
2445         (pass_sra): Convert from a global struct to a subclass of
2446         gimple_opt_pass along with...
2447         (pass_data_sra): ...new pass_data instance and...
2448         (make_pass_sra): ...new function.
2449         (pass_early_ipa_sra): Convert from a global struct to a subclass of
2450         gimple_opt_pass along with...
2451         (pass_data_early_ipa_sra): ...new pass_data instance and...
2452         (make_pass_early_ipa_sra): ...new function.
2453         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
2454         subclass of gimple_opt_pass along with...
2455         (pass_data_ccp): ...new pass_data instance and...
2456         (make_pass_ccp): ...new function.
2457         (pass_fold_builtins): Convert from a global struct to a subclass of
2458         gimple_opt_pass along with...
2459         (pass_data_fold_builtins): ...new pass_data instance and...
2460         (make_pass_fold_builtins): ...new function.
2461         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
2462         subclass of gimple_opt_pass along with...
2463         (pass_data_copy_prop): ...new pass_data instance and...
2464         (make_pass_copy_prop): ...new function.
2465         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
2466         global struct to a subclass of gimple_opt_pass along with...
2467         (pass_data_rename_ssa_copies): ...new pass_data instance and...
2468         (make_pass_rename_ssa_copies): ...new function.
2469         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
2470         subclass of gimple_opt_pass along with...
2471         (pass_data_dce): ...new pass_data instance and...
2472         (make_pass_dce): ...new function.
2473         (pass_dce_loop): Convert from a global struct to a subclass of
2474         gimple_opt_pass along with...
2475         (pass_data_dce_loop): ...new pass_data instance and...
2476         (make_pass_dce_loop): ...new function.
2477         (pass_cd_dce): Convert from a global struct to a subclass of
2478         gimple_opt_pass along with...
2479         (pass_data_cd_dce): ...new pass_data instance and...
2480         (make_pass_cd_dce): ...new function.
2481         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
2482         subclass of gimple_opt_pass along with...
2483         (pass_data_dominator): ...new pass_data instance and...
2484         (make_pass_dominator): ...new function.
2485         (pass_phi_only_cprop): Convert from a global struct to a subclass of
2486         gimple_opt_pass along with...
2487         (pass_data_phi_only_cprop): ...new pass_data instance and...
2488         (make_pass_phi_only_cprop): ...new function.
2489         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
2490         subclass of gimple_opt_pass along with...
2491         (pass_data_dse): ...new pass_data instance and...
2492         (make_pass_dse): ...new function.
2493         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
2494         a subclass of gimple_opt_pass along with...
2495         (pass_data_forwprop): ...new pass_data instance and...
2496         (make_pass_forwprop): ...new function.
2497         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
2498         struct to a subclass of gimple_opt_pass along with...
2499         (pass_data_tree_ifcombine): ...new pass_data instance and...
2500         (make_pass_tree_ifcombine): ...new function.
2501         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
2502         subclass of gimple_opt_pass along with...
2503         (pass_data_ch): ...new pass_data instance and...
2504         (make_pass_ch): ...new function.
2505         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
2506         subclass of gimple_opt_pass along with...
2507         (pass_data_tree_loop): ...new pass_data instance and...
2508         (make_pass_tree_loop): ...new function.
2509         (pass_tree_loop_init): Convert from a global struct to a subclass of
2510         gimple_opt_pass along with...
2511         (pass_data_tree_loop_init): ...new pass_data instance and...
2512         (make_pass_tree_loop_init): ...new function.
2513         (pass_lim): Convert from a global struct to a subclass of
2514         gimple_opt_pass along with...
2515         (pass_data_lim): ...new pass_data instance and...
2516         (make_pass_lim): ...new function.
2517         (pass_tree_unswitch): Convert from a global struct to a subclass of
2518         gimple_opt_pass along with...
2519         (pass_data_tree_unswitch): ...new pass_data instance and...
2520         (make_pass_tree_unswitch): ...new function.
2521         (pass_predcom): Convert from a global struct to a subclass of
2522         gimple_opt_pass along with...
2523         (pass_data_predcom): ...new pass_data instance and...
2524         (make_pass_predcom): ...new function.
2525         (pass_vectorize): Convert from a global struct to a subclass of
2526         gimple_opt_pass along with...
2527         (pass_data_vectorize): ...new pass_data instance and...
2528         (make_pass_vectorize): ...new function.
2529         (pass_graphite): Convert from a global struct to a subclass of
2530         gimple_opt_pass along with...
2531         (pass_data_graphite): ...new pass_data instance and...
2532         (make_pass_graphite): ...new function.
2533         (pass_graphite_transforms): Convert from a global struct to a subclass
2534         of gimple_opt_pass along with...
2535         (pass_data_graphite_transforms): ...new pass_data instance and...
2536         (make_pass_graphite_transforms): ...new function.
2537         (pass_check_data_deps): Convert from a global struct to a subclass of
2538         gimple_opt_pass along with...
2539         (pass_data_check_data_deps): ...new pass_data instance and...
2540         (make_pass_check_data_deps): ...new function.
2541         (pass_iv_canon): Convert from a global struct to a subclass of
2542         gimple_opt_pass along with...
2543         (pass_data_iv_canon): ...new pass_data instance and...
2544         (make_pass_iv_canon): ...new function.
2545         (pass_scev_cprop): Convert from a global struct to a subclass of
2546         gimple_opt_pass along with...
2547         (pass_data_scev_cprop): ...new pass_data instance and...
2548         (make_pass_scev_cprop): ...new function.
2549         (pass_record_bounds): Convert from a global struct to a subclass of
2550         gimple_opt_pass along with...
2551         (pass_data_record_bounds): ...new pass_data instance and...
2552         (make_pass_record_bounds): ...new function.
2553         (pass_complete_unroll): Convert from a global struct to a subclass of
2554         gimple_opt_pass along with...
2555         (pass_data_complete_unroll): ...new pass_data instance and...
2556         (make_pass_complete_unroll): ...new function.
2557         (pass_complete_unrolli): Convert from a global struct to a subclass of
2558         gimple_opt_pass along with...
2559         (pass_data_complete_unrolli): ...new pass_data instance and...
2560         (make_pass_complete_unrolli): ...new function.
2561         (pass_parallelize_loops): Convert from a global struct to a subclass
2562         of gimple_opt_pass along with...
2563         (pass_data_parallelize_loops): ...new pass_data instance and...
2564         (make_pass_parallelize_loops): ...new function.
2565         (pass_loop_prefetch): Convert from a global struct to a subclass of
2566         gimple_opt_pass along with...
2567         (pass_data_loop_prefetch): ...new pass_data instance and...
2568         (make_pass_loop_prefetch): ...new function.
2569         (pass_iv_optimize): Convert from a global struct to a subclass of
2570         gimple_opt_pass along with...
2571         (pass_data_iv_optimize): ...new pass_data instance and...
2572         (make_pass_iv_optimize): ...new function.
2573         (pass_tree_loop_done): Convert from a global struct to a subclass of
2574         gimple_opt_pass along with...
2575         (pass_data_tree_loop_done): ...new pass_data instance and...
2576         (make_pass_tree_loop_done): ...new function.
2577         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2578         struct to a subclass of gimple_opt_pass along with...
2579         (pass_data_cse_reciprocals): ...new pass_data instance and...
2580         (make_pass_cse_reciprocals): ...new function.
2581         (pass_cse_sincos): Convert from a global struct to a subclass of
2582         gimple_opt_pass along with...
2583         (pass_data_cse_sincos): ...new pass_data instance and...
2584         (make_pass_cse_sincos): ...new function.
2585         (pass_optimize_bswap): Convert from a global struct to a subclass of
2586         gimple_opt_pass along with...
2587         (pass_data_optimize_bswap): ...new pass_data instance and...
2588         (make_pass_optimize_bswap): ...new function.
2589         (pass_optimize_widening_mul): Convert from a global struct to a
2590         subclass of gimple_opt_pass along with...
2591         (pass_data_optimize_widening_mul): ...new pass_data instance and...
2592         (make_pass_optimize_widening_mul): ...new function.
2593         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2594         subclass of gimple_opt_pass along with...
2595         (pass_data_phiopt): ...new pass_data instance and...
2596         (make_pass_phiopt): ...new function.
2597         (pass_cselim): Convert from a global struct to a subclass of
2598         gimple_opt_pass along with...
2599         (pass_data_cselim): ...new pass_data instance and...
2600         (make_pass_cselim): ...new function.
2601         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2602         subclass of gimple_opt_pass along with...
2603         (pass_data_phiprop): ...new pass_data instance and...
2604         (make_pass_phiprop): ...new function.
2605         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2606         subclass of gimple_opt_pass along with...
2607         (pass_data_pre): ...new pass_data instance and...
2608         (make_pass_pre): ...new function.
2609         (pass_fre): Convert from a global struct to a subclass of
2610         gimple_opt_pass along with...
2611         (pass_data_fre): ...new pass_data instance and...
2612         (make_pass_fre): ...new function.
2613         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2614         subclass of gimple_opt_pass along with...
2615         (pass_data_reassoc): ...new pass_data instance and...
2616         (make_pass_reassoc): ...new function.
2617         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2618         subclass of gimple_opt_pass along with...
2619         (pass_data_sink_code): ...new pass_data instance and...
2620         (make_pass_sink_code): ...new function.
2621         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2622         subclass of gimple_opt_pass along with...
2623         (pass_data_strlen): ...new pass_data instance and...
2624         (make_pass_strlen): ...new function.
2625         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2626         struct to a subclass of gimple_opt_pass along with...
2627         (pass_data_build_alias): ...new pass_data instance and...
2628         (make_pass_build_alias): ...new function.
2629         (pass_build_ealias): Convert from a global struct to a subclass of
2630         gimple_opt_pass along with...
2631         (pass_data_build_ealias): ...new pass_data instance and...
2632         (make_pass_build_ealias): ...new function.
2633         (pass_ipa_pta): Convert from a global struct to a subclass of
2634         simple_ipa_opt_pass along with...
2635         (pass_data_ipa_pta): ...new pass_data instance and...
2636         (make_pass_ipa_pta): ...new function.
2637         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2638         subclass of gimple_opt_pass along with...
2639         (pass_data_uncprop): ...new pass_data instance and...
2640         (make_pass_uncprop): ...new function.
2641         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2642         global struct to a subclass of gimple_opt_pass along with...
2643         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2644         (make_pass_late_warn_uninitialized): ...new function.
2645         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2646         to a subclass of gimple_opt_pass along with...
2647         (pass_data_init_datastructures): ...new pass_data instance and...
2648         (make_pass_init_datastructures): ...new function.
2649         (pass_early_warn_uninitialized): Convert from a global struct to a
2650         subclass of gimple_opt_pass along with...
2651         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2652         (make_pass_early_warn_uninitialized): ...new function.
2653         (pass_update_address_taken): Convert from a global struct to a
2654         subclass of gimple_opt_pass along with...
2655         (pass_data_update_address_taken): ...new pass_data instance and...
2656         (make_pass_update_address_taken): ...new function.
2657         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2658         struct to a subclass of gimple_opt_pass along with...
2659         (pass_data_release_ssa_names): ...new pass_data instance and...
2660         (make_pass_release_ssa_names): ...new function.
2661         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2662         subclass of gimple_opt_pass along with...
2663         (pass_data_stdarg): ...new pass_data instance and...
2664         (make_pass_stdarg): ...new function.
2665         * tree-switch-conversion.c (pass_convert_switch): Convert from a
2666         global struct to a subclass of gimple_opt_pass along with...
2667         (pass_data_convert_switch): ...new pass_data instance and...
2668         (make_pass_convert_switch): ...new function.
2669         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2670         to a subclass of gimple_opt_pass along with...
2671         (pass_data_tail_recursion): ...new pass_data instance and...
2672         (make_pass_tail_recursion): ...new function.
2673         (pass_tail_calls): Convert from a global struct to a subclass of
2674         gimple_opt_pass along with...
2675         (pass_data_tail_calls): ...new pass_data instance and...
2676         (make_pass_tail_calls): ...new function.
2677         * tree-vect-generic.c (pass_lower_vector): Convert from a global
2678         struct to a subclass of gimple_opt_pass along with...
2679         (pass_data_lower_vector): ...new pass_data instance and...
2680         (make_pass_lower_vector): ...new function.
2681         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2682         gimple_opt_pass along with...
2683         (pass_data_lower_vector_ssa): ...new pass_data instance and...
2684         (make_pass_lower_vector_ssa): ...new function.
2685         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2686         to a subclass of gimple_opt_pass along with...
2687         (pass_data_slp_vectorize): ...new pass_data instance and...
2688         (make_pass_slp_vectorize): ...new function.
2689         (pass_ipa_increase_alignment): Convert from a global struct to a
2690         subclass of simple_ipa_opt_pass along with...
2691         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2692         (make_pass_ipa_increase_alignment): ...new function.
2693         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2694         gimple_opt_pass along with...
2695         (pass_data_vrp): ...new pass_data instance and...
2696         (make_pass_vrp): ...new function.
2697         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2698         subclass of simple_ipa_opt_pass along with...
2699         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2700         (make_pass_ipa_free_lang_data): ...new function.
2701         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2702         gimple_opt_pass along with...
2703         (pass_data_tsan): ...new pass_data instance and...
2704         (make_pass_tsan): ...new function.
2705         (pass_tsan_O0): Convert from a global struct to a subclass of
2706         gimple_opt_pass along with...
2707         (pass_data_tsan_O0): ...new pass_data instance and...
2708         (make_pass_tsan_O0): ...new function.
2709         * var-tracking.c (pass_variable_tracking): Convert from a global
2710         struct to a subclass of rtl_opt_pass along with...
2711         (pass_data_variable_tracking): ...new pass_data instance and...
2712         (make_pass_variable_tracking): ...new function.
2713         * web.c (pass_web): Convert from a global struct to a subclass of
2714         rtl_opt_pass along with...
2715         (pass_data_web): ...new pass_data instance and...
2716         (make_pass_web): ...new function.
2717         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
2718         declaration with that of...
2719         (make_pass_mode_switch_use): ...new function.
2720         (pass_resolve_sw_modes): Replace declaration with that of...
2721         (make_pass_resolve_sw_modes): ...new function.
2722         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
2723         from a global struct to a subclass of rtl_opt_pass along with...
2724         (pass_data_mode_switch_use): ...new pass_data instance and...
2725         (make_pass_mode_switch_use): ...new function.
2726         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
2727         from a global struct to a subclass of rtl_opt_pass along with...
2728         (pass_data_resolve_sw_modes): ...new pass_data instance and...
2729         (make_pass_resolve_sw_modes): ...new function.
2730         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
2731         struct to a subclass of rtl_opt_pass along with...
2732         (pass_data_insert_vzeroupper): ...new pass_data instance and...
2733         (make_pass_insert_vzeroupper): ...new function.
2734         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
2735         global struct to a subclass of rtl_opt_pass along with...
2736         (pass_data_work_around_errata): ...new pass_data instance and...
2737         (make_pass_work_around_errata): ...new function.
2738         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
2739         struct to a subclass of rtl_opt_pass along with...
2740         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
2741         (make_pass_mips_machine_reorg2): ...new function.
2743 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2745         * passes.c (pass_manager::operator new): New.
2747 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2749         Handwritten part of conversion of passes to C++ classes.
2751         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
2752         (toplev.o): Add dep on PASS_MANAGER_H.
2753         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
2754         of early local pases to reflect this moving from a global to a
2755         member of gcc::pass_manager.
2756         (cgraph_add_new_function): Likewise.
2757         * lto-cgraph.c (lto_output_node): Update for conversion of
2758         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
2759         * passes.c (opt_pass::clone): New.
2760         (opt_pass::gate): New.
2761         (opt_pass::execute): New.
2762         (opt_pass::opt_pass): New.
2763         (pass_manager::execute_early_local_passes): New.
2764         (pass_manager::execute_pass_mode_switching): new.
2765         (finish_optimization_passes): Convert to...
2766         (pass_manager::finish_optimization_passes): ...this.
2767         (finish_optimization_passes): Update for conversion of passes to
2768         C++ classes.
2769         (register_dump_files_1): Use has_gate since we cannot portably
2770         check a vtable entry against NULL.
2771         (dump_one_pass): Likewise.
2772         (ipa_write_summaries_2): Likewise.
2773         (ipa_write_optimization_summaries_1): Likewise.
2774         (ipa_read_summaries_1): Likewise.
2775         (ipa_read_optimization_summaries_1): Likewise.
2776         (execute_ipa_stmt_fixups): Likewise.
2777         (pass_manager::pass_manager): Rewrite pass-creation, invoking
2778         pass-creation functions rather than wiring up globals, and
2779         storing the results in fields of pass_manager generated using
2780         pass-instances.def.
2781         (pass_manager::dump_profile_report): Update for conversion of
2782         passes to C++ classes.
2783         (pass_manager::execute_ipa_summary_passes): Likewise.
2784         (execute_one_ipa_transform_pass): Likewise.
2785         (execute_one_pass): Use has_gate and has_execute since we cannot
2786         portably check a vtable entry against NULL.
2787         * pass_manager.h (pass_manager::finish_optimization_passes): New.
2788         (pass_manager): Use pass-instances.def to add fields for the
2789         various pass instances.
2790         * toplev.c (finalize): Update for move of
2791         finish_optimization_passes to a method of gcc::pass_manager.
2792         * toplev.h (finish_optimization_passes): Move to method of class
2793         pass_manager.
2794         * tree-pass.h (struct pass_data): New.
2795         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
2796         (opt_pass::gate): Convert to virtual function.
2797         (opt_pass::~opt_pass): New.
2798         (opt_pass::clone): New.
2799         (opt_pass::execute): Convert to virtual function.
2800         (opt_pass::opt_pass): New.
2801         (opt_pass::ctxt_): new.
2802         (gimple_opt_pass): Convert to subclass of opt_pass.
2803         (gimple_opt_pass::gimple_opt_pass): New.
2804         (rtl_opt_pass): Convert to subclass of opt_pass.
2805         (rtl_opt_pass::rtl_opt_pass): New.
2806         (ipa_opt_pass_d): Convert to subclass of opt_pass.
2807         (ipa_opt_pass_d::ipa_opt_pass_d): New.
2808         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2809         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2810         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2811         invocation of pass_mode_switching to reflect this moving from a
2812         global to a member of gcc::pass_manager.
2813         (ix86_option_override): Rework how pass_insert_vzeroupper is
2814         added to the pass_manager to reflect autogenerated changes.
2815         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2817 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
2819         PR rtl-optimization/57708
2820         * recog.c (peep2_find_free_register): Validate all regs in a
2821         multi-reg mode.
2823 2013-08-05  Jan Hubicka  <jh@suse.cz>
2825         PR lto/57602
2826         * cgraph.c (verify_cgraph_node): Accept local flags from other
2827         partitions.
2828         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2829         (function_and_variable_visibility): Likewise.
2830         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2832 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2834         * graph.c (init_graph_slim_pretty_print): Remove.
2835         (print_graph_cfg): Do not call it.  Use local pretty printer.
2836         (start_graph_dump): Likewise.
2838 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2840         * gimple-pretty-print.c (buffer): Remove.
2841         (initialized): Likewise.
2842         (maybe_init_pretty_print): Likewise.
2843         (print_gimple_stmt): Do not call it.  Use non-static local
2844         pretty_printer variable.
2845         (print_gimple_expr): Likewise.
2846         (print_gimple_seq): Likewise.
2847         (gimple_dump_bb): Likewise.
2849 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2851         * asan.c (asan_pp): Remove.
2852         (asan_pp_initialized): Likewise.
2853         (asan_pp_initialize): Likewise.
2854         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
2855         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
2856         (asan_add_global): Likewise.
2858 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2860         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
2861         * diagnostic.h (diagnostic_flush_buffer): Adjust.
2862         * pretty-print.c (pp_formatted_text_data): Likewise.
2863         (pp_indent): Rename from pp_base_indent.
2864         (pp_format): Rename from pp_base_format.
2865         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2866         (pp_format_verbatim): Rename from pp_base_format_verbatim.
2867         (pp_flush): Rename from pp_base_flush.
2868         (pp_set_line_maximum_length): Rename from
2869         pp_base_set_line_maximum_length.
2870         (pp_clear_output_area): Rename from pp_base_clear_output_area.
2871         (pp_set_prefix): Rename from pp_base_set_prefix.
2872         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2873         (pp_emit_prefix): Rename from pp_base_emit_prefix.
2874         (pp_append_text): Rename from pp_base_append_text.
2875         (pp_formatted_text): Rename from pp_base_formatted_text.
2876         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2877         (pp_remaining_character_count_for_line): Rename from
2878         pp_base_remaining_character_count_for_line.
2879         (pp_newline): Rename from pp_base_newline.
2880         (pp_character): Rename from pp_base_character.
2881         (pp_string): Rename from pp_base_string.
2882         (pp_maybe_space): Rename from pp_base_maybe_space.
2883         * asan.c (asan_pp_string): Adjust.
2884         (asan_emit_stack_protection): Likewise.
2885         (asan_add_global): Likewise.
2886         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2887         * tree-mudflap.c (mf_varname_tree): Likewise.
2888         * tree-pretty-print.c (pp_tree_identifier): Rename from
2889         pp_base_tree_identifier.
2890         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2891         Declare as function.
2893 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2895         * pretty-print.h (pp_bar_bar): New.
2896         (pp_ampersand_ampersand): Likewise.
2897         (pp_less_equal): Likewise.
2898         (pp_greater_equal): Likewise.
2899         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2900         printer functions instead of pp_string or operators and punctuators.
2901         (dump_gimple_call): Likewise.
2902         (dump_gimple_omp_for): Likewise.
2903         (dump_gimple_transaction): Likewise.
2904         (dump_gimple_phi): Likewise.
2905         (pp_gimple_stmt_1): Likewise.
2906         * sched-vis.c (print_insn): Likewise.
2907         * tree-mudflap.c (mf_varname_tree): Likewise.
2908         * tree-pretty-print.c (dump_block_node): Likewise.
2909         (dump_generic_node): Likewise.
2911 2013-08-02  Jan Hubicka  <jh@suse.cz>
2913         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2914         boundaries.
2915         * lto-streamer-out.c (tree_is_indexable): Results decls and
2916         parm decls are not indexable.
2917         (DFS_write_tree_body): Do not follow args and results.
2918         (hash_tree): Likewise.
2919         (output_functions): Rearrange so struct function is needed
2920         only when real body is output; be able to also ouptut abstract
2921         functions; output DECL_ARGUMENTS and DECL_RESULT.
2922         (lto_output): When not in WPA, ale store abstract functions.
2923         (write_symbol): Do not care about RESULT_DECL.
2924         (output_symbol_p): Handle correctly sbtract decls.
2925         * lto-streamer-in.c (input_function): Rearrange so struct
2926         function can be NULL at entry; allow streaming of
2927         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2928         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2929         sanity check during LTO.
2930         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2931         RESULT_DECl and DECL_ARGUMENTS.
2932         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2933         Likewise.
2935 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2937         * pretty-print.h (pp_underscore): New.
2938         (pp_comma): Tidy.
2939         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2940         printer functions instead of pp_character.
2941         (dump_binary_rhs): Likewise.
2942         (dump_ternary_rhs): Likewise.
2943         (dump_gimple_call_args): Likewise.
2944         (pp_points_to_solution): Likewise.
2945         (dump_gimple_call): Likewise.
2946         (dump_gimple_switch): Likewise.
2947         (dump_gimple_cond): Likewise.
2948         (dump_gimple_bind): Likewise.
2949         (dump_gimple_try): Likewise.
2950         (dump_gimple_omp_for): Likewise.
2951         (dump_gimple_omp_continue): Likewise.
2952         (dump_gimple_omp_single): Likewise.
2953         (dump_gimple_omp_sections): Likewise.
2954         (dump_gimple_omp_block): Likewise.
2955         (dump_gimple_omp_critical): Likewise.
2956         (dump_gimple_transaction): Likewise.
2957         (dump_gimple_asm): Likewise.
2958         (dump_gimple_phi): Likewise.
2959         (dump_gimple_omp_parallel): Likewise.
2960         (dump_gimple_omp_task): Likewise.
2961         (dump_gimple_omp_atomic_load): Likewise.
2962         (dump_gimple_omp_atomic_store): Likewise.
2963         (dump_gimple_mem_ops): Likewise.
2964         (pp_gimple_stmt_1): Likewise.
2965         (pp_cfg_jump): Likewise.
2966         (dump_implicit_edges): Likewise.
2967         (gimple_dump_bb_for_graph): Likewise.
2968         * graph.c (draw_cfg_node): Likewise.
2969         * langhooks.c (lhd_print_error_function): Likewise.
2970         * sched-vis.c (print_exp): Likewise.
2971         (print_value): Likewise.
2972         (print_pattern): Likewise.
2973         (print_insn): Likewise.
2974         (rtl_dump_bb_for_graph): Likewise.
2975         * tree-pretty-print.c (dump_function_declaration): Likewise.
2976         (dump_array_domain): Likewise.
2977         (dump_omp_clause): Likewise.
2978         (dump_location): Likewise.
2979         (dump_generic_node): Likewise.
2980         (print_struct_decl): Likewise.
2981         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2983 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2985         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2986         candidate table when replacing a candidate statement.
2987         (replace_rhs_if_not_dup): Likewise.
2988         (replace_one_candidate): Likewise.
2990 2013-08-02  Jan Hubicka  <jh@suse.cz>
2991             Martin Liska  <marxin.liska@gmail.com>
2993         * cgraphunit.c (add_new_function): Fix logic when adding from
2994         late IPA pass.
2995         (assemble_thunk): Rename to ...
2996         (expand_thunk); .. this one; export; get it working with
2997         general functions; make produced gimple valid.
2998         * cgraph.h (expand_thunk): Declare.
3000 2013-08-02  Jan Hubicka  <jh@suse.cz>
3002         * ipa-cp.c (gather_context_independent_values): Use
3003         ipa_get_param_move_cost.
3004         (get_replacement_map): Remove PARAM; move parameter folding
3005         into tree-inline.c
3006         (create_specialized_node): Update.
3007         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3008         assert that we have gimple body; update move_cost.
3009         (count_formal_params): Assert that we have gimple body.
3010         (ipa_dump_param): New function.
3011         (ipa_alloc_node_params): Break out from ...
3012         (ipa_initialize_node_params): ... here.
3013         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3014         (ipa_write_node_info): Stream move costs.
3015         (ipa_read_node_info): Read move costs.
3016         (ipa_update_after_lto_read): Do not recompute node params.
3017         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3018         (ipa_get_param): Check we are not in WPA.
3019         (ipa_get_param_move_cost): New.
3020         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3021         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3022         parm numbers to be present.
3024 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3026         PR rtl-optimization/58048
3027         * lra-constraints.c (process_alt_operands): Don't check asm
3028         operand on register.
3030 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
3032         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3033         the implied StoreLoad barrier for atomic operations if before.
3035 2013-08-02  Jan Hubicka  <jh@suse.cz>
3036             Martin Liska  <marxin.liska@gmail.com>
3038         * cgraph.c (cgraph_function_body_availability): Do not check
3039         cgraph flags.
3040         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3041         symtab_node_availability): Declare.
3042         * ipa.c (can_replace_by_local_alias): New.
3043         (function_and_variable_visibility): Use it.
3044         * symtab.c (symtab_for_node_and_aliases,
3045         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3047 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3049         PR rtl-optimization/57963
3050         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3051         (lra_constraints): Use them.
3053 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
3055         * config/arm/types.md (define_attr "type"): Add "load_acq"
3056         and "store_rel".
3057         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3058         changes.
3059         (cortex_a53_store1): Likewise.
3061 2013-08-01  Jan Hubicka  <jh@suse.cz>
3063         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3064         partitions are not needed.
3066 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3068         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3069         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3070         MAYBE_NON_Q_CLASS_P where appropriate.
3072 2013-08-01  Jan Hubicka  <jh@suse.cz>
3074         * cgraph.h (release_function_body): Declare.
3075         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3076         of unused delcarations.
3078 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3080         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3081         RTL form when subtracting a constant.
3083 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3085         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3086         Generate canonical plus rtx with negated immediate instead of minus
3087         where appropriate.
3088         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3090 2013-08-01  Jan Hubicka  <jh@suse.cz>
3092         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3093         (cgraph_release_function_body): Likewise.
3094         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3095         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3096         to used_as_abstract_origin.
3097         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3098         symbols used as abstract origins.
3099         * cgraphunit.c (analyze_functions): Update.
3100         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3101         used_as_abstract_origin.
3102         * tree-inline.c (tree_function_versioning): Update
3103         used_as_abstract_origin; be ready for DECL_RESULT and
3104         DECL_ARGUMENTS to be NULL.
3106         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3107         for abstract functions.
3108         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3109         real symbols.
3111 2013-08-01  Jan Hubicka  <jh@suse.cz>
3113         * profile.c (compute_value_histograms): Fix thinko.
3115 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3117         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
3118         aarch-common-protos.h to extra_headers.
3119         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3120         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3121         * config/aarch64/t-aarch64 (aarch-common.o): Define.
3123 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3125         * config/aarch64/aarch64.md (define_attr "type"): Delete.
3126         Include "../arm/types.md".  Define "type" attribute for all patterns.
3127         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3128         attribute changes.
3130 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
3132         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3133         to support power8 load fusion.
3134         (fusion_gpr_mem_load): Likewise.
3136         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3138         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3139         declarations for power8 load fusion.
3140         (emit_fusion_gpr_load): Likewise.
3142         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3143         tuning for power8, turn on fusion mode by default.  Turn on sign
3144         extending fusion mode if normal fusion mode is on, and we are at
3145         -O2 or -O3.
3146         (fusion_gpr_load_p): New function, return true if we can fuse an
3147         addis instruction with a dependent load to a GPR.
3148         (emit_fusion_gpr_load): Emit the instructions for power8 load
3149         fusion to GPRs.
3151         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3152         (VSX load fusion peepholes): New peepholes to fuse together an
3153         addi instruction with a VSX load instruction.
3155         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3156         peepholes to fuse an addis instruction with a load to a GPR base
3157         register.  If we are supporting sign extending fusions, convert
3158         sign extending loads to zero extending loads and add an explicit
3159         sign extension.
3161 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3163         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
3164         aarch-common-protos.h to extra_headers.
3165         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3166         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3167         (arm_early_store_addr_dep): Likewise.
3168         (arm_no_early_alu_shift_dep): Likewise.
3169         (arm_no_early_alu_shift_value_dep): Likewise.
3170         (arm_no_early_mul_dep): Likewise.
3171         (arm_no_early_store_addr_dep): Likewise.
3172         (arm_mac_accumulator_is_mul_result): Likewise.
3173         (arm_mac_accumulator_is_result): Likewise.
3174         * config/arm/aarch-common.c: ... here.  New file.
3175         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3176         here to ...
3177         (arm_early_store_addr_dep): Likewise.
3178         (arm_no_early_alu_shift_dep): Likewise.
3179         (arm_no_early_alu_shift_value_dep): Likewise.
3180         (arm_no_early_mul_dep): Likewise.
3181         (arm_no_early_store_addr_dep): Likewise.
3182         (arm_mac_accumulator_is_mul_result): Likewise.
3183         (arm_mac_accumulator_is_result): Likewise.
3184         * config/arm/aarch-common-protos.h: ... here.  New file.
3185         * config/arm/t-arm (aarch-common.o): Define.
3187 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3189         * config/arm/arm.md: Include new file "types.md".
3190         (define_attr "type"): Move from here to ...
3191         (define_attr "mul32"): Likewise.
3192         (define_attr "mul64"): Likewise.
3193         * config/arm/types.md: ... here.  New file.
3195 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
3197         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3198         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3200 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3202         * gen-pass-instances.awk: Fix offset of substr().
3204 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
3206         * Makefile.in (pass-instances.def): New.
3207         (passes.o): Replace dependency on passes.def with one on
3208         pass-instances.def
3210         * gen-pass-instances.awk: New.
3212         * passes.c (pass_manager::pass_manager): Use pass-instances.def
3213         rather than passes.def, updating local definition of NEXT_PASS
3214         macro to add an extra NUM parameter (currently unused).
3216 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
3218         * Makefile.in (PASS_MANAGER_H): New.
3219         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3220         (passes.o): Likewise.
3221         (statistics.o): Likewise.
3222         (cgraphunit.o): Likewise.
3223         (context.o): Depend on PASS_MANAGER_H.
3225         * pass_manager.h: New.
3227         * cgraphunit.c (cgraph_add_new_function): Update for moves
3228         of globals to fields of pass_manager.
3229         (analyze_function): Likewise.
3230         (expand_function): Likewise.
3231         (ipa_passes): Likewise.
3232         (compile): Likewise.
3234         * context.c (context::context): New.
3235         * context.h  (context::context): New.
3236         (context::get_passes): New.
3237         (context::passes_): New.
3239         * lto-cgraph.c (input_node): Update for moves of globals to
3240         fields of pass_manager.
3242         * passes.c (all_passes): Remove, in favor of a field of the
3243         same name within the new class pass_manager.
3244         (all_small_ipa_passes): Likewise.
3245         (all_lowering_passes): Likewise.
3246         (all_regular_ipa_passes): Likewise.
3247         (all_late_ipa_passes): Likewise.
3248         (all_lto_gen_passes): Likewise.
3249         (passes_by_id): Likewise.
3250         (passes_by_id_size): Likewise.
3251         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3252         the new class pass_manager.
3253         (set_pass_for_id): Convert to...
3254         (pass_manager::set_pass_for_id): ...method.
3255         (get_pass_for_id): Convert to...
3256         (pass_manager::get_pass_for_id): ...method.
3257         (register_one_dump_file): Move body of implementation into...
3258         (pass_manager::register_one_dump_file): ...here.
3259         (register_dump_files_1): Convert to...
3260         (pass_manager::register_dump_files_1): ...method.
3261         (register_dump_files): Convert to...
3262         (pass_manager::register_dump_files): ...method.
3263         (create_pass_tab): Update for moves of globals to fields of
3264         pass_manager.
3265         (dump_passes): Move body of implementation into...
3266         (pass_manager::dump_passes): ...here.
3267         (register_pass): Move body of implementation into...
3268         (pass_manager::register_pass): ...here.
3269         (init_optimization_passes): Convert into...
3270         (pass_manager::pass_manager): ...constructor for new
3271         pass_manager class, and initialize the pass_lists array.
3272         (check_profile_consistency): Update for moves of globals to
3273         fields of pass_manager.
3274         (dump_profile_report): Move body of implementation into...
3275         (pass_manager::dump_profile_report): ...here.
3276         (ipa_write_summaries_1): Update for moves of pass lists from
3277         being globals to fields of pass_manager.
3278         (ipa_write_optimization_summaries): Likewise.
3279         (ipa_read_summaries):  Likewise.
3280         (ipa_read_optimization_summaries): Likewise.
3281         (execute_all_ipa_stmt_fixups): Likewise.
3283         * statistics.c (statistics_fini): Update for moves of globals to
3284         fields of pass_manager.
3286         * toplev.c (general_init): Replace call to
3287         init_optimization_passes with construction of the pass_manager
3288         instance.
3290         * tree-pass.h (all_passes): Remove, in favor of a field of the
3291         same name within the new class pass_manager.
3292         (all_small_ipa_passes): Likewise.
3293         (all_lowering_passes): Likewise.
3294         (all_regular_ipa_passes): Likewise.
3295         (all_lto_gen_passes): Likewise.
3296         (all_late_ipa_passes): Likewise.
3297         (passes_by_id): Likewise.
3298         (passes_by_id_size): Likewise.
3299         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3300         the new class pass_manager.
3301         (get_pass_for_id): Remove.
3303 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3305         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
3306         configs.
3308 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3310         * arm.md (mulhi3): New expand pattern.
3312 2013-07-30  Jan Hubicka  <jh@suse.cz>
3313             Martin Liska  <marxin.liska@gmail.com>
3315         * profile.c (compute_value_histograms): Do not ICE when
3316         there is mismatch only on some counters.
3318 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3320         PR rtl-optimization/57637
3321         * function.c (move_insn_for_shrink_wrap): Also check the
3322         GEN set of the LIVE problem for the liveness analysis
3323         if it exists, otherwise give up.
3325 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3327         PR tree-optimization/57993
3328         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
3329         replaced statement in the candidate table.
3330         (phi_add_costs): Return infinite cost when the hidden basis does
3331         not dominate all phis on which the candidate is dependent.
3332         (replace_one_candidate): Record replaced statement in the
3333         candidate table.
3335 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
3337         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
3338         (ashlv2si3): New expander.
3339         (*ashlv2si3_i): New define_insn_and_split.
3340         * predicates.md (float_operation): Allow patterns with three
3341         basic sub-patterns.
3343         PR rtl-optimization/58021
3344         * mode-switching.c (create_pre_exit): Always split off preceding
3345         insns if we are not at the basic block head.
3347 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
3349         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
3350         (UCLIBC_DYNAMIC_LINKER): New macro.
3351         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
3352         `-mnan=2008'.
3353         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
3354         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
3355         `-mnan=2008'.
3356         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
3357         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
3358         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
3359         for SF and DF modes.  Use ieee_quad_format for TF mode.
3360         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
3361         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
3362         (mips_option_override): Handle `-mnan=legacy'.
3363         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
3364         `-mabs=2008' and `-mnan=2008'.
3365         (OPTION_DEFAULT_SPECS): Add "nan" default.
3366         (ASM_SPEC): Handle `-mnan='.
3367         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
3368         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
3369         comment accordingly.
3370         (neg<mode>2): Likewise.
3371         * config/mips/mips.opt (mabs, mnan): New options.
3372         * doc/install.texi (Configuration): Document `--with-nan=' option.
3373         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
3374         `-mnan=' options.
3375         (MIPS Options): Document them.
3376         * config.gcc <mips*-*-*>: Handle `--with-nan='.
3377         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
3378         * configure: Regenerate.
3379         * config.in: Regenerate.
3381 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3383         * config/i386/i386.md (float post-reload splitters): Do not check
3384         for subregs of SSE registers.
3386 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3387             H.J. Lu  <hongjiu.lu@intel.com>
3389         PR target/57954
3390         PR target/57988
3391         * config/i386/i386.md (post-reload splitter
3392         to avoid partial SSE reg dependency stalls): New pattern.
3394 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3396         * config/s390/s390.md ("movcc"): Swap load and store instructions.
3398 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
3400         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
3401         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
3403 2013-07-26  Cary Coutant  <ccoutant@google.com>
3405         * dwarf2out.c (die_checksum_ordered): Don't include template
3406         instantiations in signature.
3407         (is_template_parameter): New function.
3408         (is_template_instantiation): New function.
3409         (generate_skeleton_bottom_up): Don't include template instantiations
3410         in type unit DIE.
3411         (generate_skeleton): Likewise.
3412         (break_out_comdat_types): Move recursive call to break out nested
3413         types earlier.
3414         (prune_unused_types_mark_generic_parms_dies): Call
3415         is_template_parameter.
3417 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
3419         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
3420         uses vector registers.
3421         * config/aarch64/iterators.md: Add attributes rtn and vas.
3423 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3424             Richard Earnshaw  <richard.earnshaw@arm.com>
3426         * combine.c (simplify_comparison): Re-canonicalize operands
3427         where appropriate.
3428         * config/arm/arm.md (movcond_addsi): New splitter.
3430 2013-07-25  Sterling Augustine  <saugustine@google.com>
3432         * dwarf2out.c (size_of_pubnames): Move code to...
3433         (include_pubname_in_output): ...here.  New.
3434         (want_pubnames): Rearrange.
3435         (output_pubnames): Call include_pubname_in_output.  Move assertion.
3437 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3439         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
3441 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3443         PR target/38836
3444         * doc/extend.texi: Remove obsolete builtins. Fix
3445         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
3447 2013-07-25  Jan Hubicka  <jh@suse.cz>
3449         * cgraph.c (release_function_body): Break out from ...
3450         (cgraph_release_function_body): ... this one; also release DECL_RESULT
3451         and DECL_ARGUMENTS.
3452         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
3453         old_tree in the map.
3454         (create_specialized_node): Update.
3455         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
3456         into index.
3457         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
3458         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
3459         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
3460         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
3461         DECL_RESULT.
3463 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3465         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
3466         addsi3_carryin_alt2_<optab>): Correct output template.
3468 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3470         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
3471         Adjust for arm_restrict_it.
3472         Remove trailing whitespace.
3474 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
3476         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
3477         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
3479         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
3481 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
3483         PR rtl-optimization/57960
3484         * lra-constraints.c (process_alt_operands): Use the right mode
3485         when checking strict_low.
3487 2013-07-25  Jan Hubicka  <jh@suse.cz>
3489         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
3490         * cgraph.c (cgraph_remove_node): Do not release function body
3491         when in cgraph streaming.
3492         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
3493         in other partitions are not considered reachable; fix handling of
3494         clones.
3496 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3498         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
3500 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3502         PR target/19599
3503         PR target/57731
3504         PR target/57837
3505         * config/arm/arm.md ("*sibcall_insn): Replace use of
3506         Ss with US. Adjust output for v5 and v4t.
3507         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
3509         * config/arm/constraints.md ("Ss"): Rename to US.
3511 2013-07-25  Terry Guo  <terry.guo@arm.com>
3513         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
3514         shift_add/shift_sub0/shift_sub1 RTXs.
3516 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
3517             Anton Blanchard <anton@au1.ibm.com>
3519         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
3520         (altivec_vpks<VI_char>ss): Likewise.
3521         (altivec_vpks<VI_char>us): Likewise.
3522         (altivec_vpku<VI_char>us): Likewise.
3523         (altivec_vpku<VI_char>um): Likewise.
3525 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
3527         Introduce context class.
3529         * Makefile.in (CONTEXT_H): New.
3530         (OBJS): Add context.o.
3531         (toplev.o): Add CONTEXT_H to dependencies.
3532         (context.o): New.
3534         * toplev.c (general_init): Create the singleton gcc::context instance.
3536         * context.c: New.
3538         * context.h: New.
3540 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
3542         PR rtl-optimization/57968
3543         * mode-switching.c (create_pre_exit): Allow instructions that
3544         don't set a return register to need a non-exit mode.
3546 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3547             Anton Blanchard <anton@au1.ibm.com>
3549         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3550         operands to vperm for little endian.
3551         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3552         of lvsl to create the control mask for a vperm for little endian.
3554 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3555             Anton Blanchard <anton@au1.ibm.com>
3557         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3558         two operands for little-endian.
3560 2013-07-23  Steve Ellcey  <sellcey@mips.com>
3562         * config/mips/mips.c (mips_case_values_threshold): New.
3563         (TARGET_CASE_VALUES_THRESHOLD): Define.
3565 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3566             Anton Blanchard <anton@au1.ibm.com>
3568         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3569         selection of field for vector splat in little endian mode.
3571 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
3573         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3574         expanders to rs6000.md.
3575         (ior<mode>3): Likewise.
3576         (and<mode>3): Likewise.
3577         (one_cmpl<mode>2): Likewise.
3578         (nor<mode>3): Likewise.
3579         (andc<mode>3): Likewise.
3580         (eqv<mode>3): Likewise.
3581         (nand<mode>3): Likewise.
3582         (orc<mode>3): Likewise.
3584         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3585         declaration.
3587         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3588         to split multi-word logical operations.
3589         (rs6000_split_logical_di): Likewise.
3590         (rs6000_split_logical): Likewise.
3592         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3593         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3594         and allow TImode operations in 32-bit.
3595         (vsx_and<mode>3_64bit): Likewise.
3596         (vsx_ior<mode>3_32bit): Likewise.
3597         (vsx_ior<mode>3_64bit): Likewise.
3598         (vsx_xor<mode>3_32bit): Likewise.
3599         (vsx_xor<mode>3_64bit): Likewise.
3600         (vsx_one_cmpl<mode>2_32bit): Likewise.
3601         (vsx_one_cmpl<mode>2_64bit): Likewise.
3602         (vsx_nor<mode>3_32bit): Likewise.
3603         (vsx_nor<mode>3_64bit): Likewise.
3604         (vsx_andc<mode>3_32bit): Likewise.
3605         (vsx_andc<mode>3_64bit): Likewise.
3606         (vsx_eqv<mode>3_32bit): Likewise.
3607         (vsx_eqv<mode>3_64bit): Likewise.
3608         (vsx_nand<mode>3_32bit): Likewise.
3609         (vsx_nand<mode>3_64bit): Likewise.
3610         (vsx_orc<mode>3_32bit): Likewise.
3611         (vsx_orc<mode>3_64bit): Likewise.
3613         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3614         logical types in GPRs.
3616         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3617         logical insns to rs6000.md, and allow TImode operations in
3618         32-bit.
3619         (altivec_ior<mode>3): Likewise.
3620         (altivec_xor<mode>3): Likewise.
3621         (altivec_one_cmpl<mode>2): Likewise.
3622         (altivec_nor<mode>3): Likewise.
3623         (altivec_andc<mode>3): Likewise.
3625         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3626         attributes for moving the 128-bit logical operations into
3627         rs6000.md.
3628         (BOOL_REGS_OUTPUT): Likewise.
3629         (BOOL_REGS_OP1): Likewise.
3630         (BOOL_REGS_OP2): Likewise.
3631         (BOOL_REGS_UNARY): Likewise.
3632         (BOOL_REGS_AND_CR0): Likewise.
3633         (one_cmpl<mode>2): Add support for DI logical operations on
3634         32-bit, splitting the operations to 32-bit.
3635         (anddi3): Likewise.
3636         (iordi3): Likewise.
3637         (xordi3): Likewise.
3638         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3639         changes to combine the 32/64-bit code, allow logical operations on
3640         TI mode in 32-bit, and to use similar match_operator patterns like
3641         scalar mode uses.  Combine the Altivec and VSX code for logical
3642         operations, and move it here.
3643         (ior<mode>3, 128-bit types): Likewise.
3644         (xor<mode>3, 128-bit types): Likewise.
3645         (one_cmpl<mode>3, 128-bit types): Likewise.
3646         (nor<mode>3, 128-bit types): Likewise.
3647         (andc<mode>3, 128-bit types): Likewise.
3648         (eqv<mode>3, 128-bit types): Likewise.
3649         (nand<mode>3, 128-bit types): Likewise.
3650         (orc<mode>3, 128-bit types): Likewise.
3651         (and<mode>3_internal): Likewise.
3652         (bool<mode>3_internal): Likewise.
3653         (boolc<mode>3_internal1): Likewise.
3654         (boolc<mode>3_internal2): Likewise.
3655         (boolcc<mode>3_internal1): Likewise.
3656         (boolcc<mode>3_internal2): Likewise.
3657         (eqv<mode>3_internal1): Likewise.
3658         (eqv<mode>3_internal2): Likewise.
3659         (one_cmpl1<mode>3_internal): Likewise.
3661 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3663         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3664         Rename flag_stack_usage to flag_stack_usage_info.
3666 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3668         * config/microblaze/sync.md: New file.
3669         * config/microblaze/microblaze.md: Include sync.md
3670         * config/microblaze/microblaze.c: Add print_operand 'y'.
3671         * config/microblaze/constraints.md: Add memory_contraint
3672         'Q' which is a single register.
3674 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
3676         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3678 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3680         * reload.c (find_reloads): Exit loop once we find this operand
3681         cannot be reloaded somehow for this alternative.
3683         * reload.c (find_reloads): Exit loop once we find a hard register.
3685         * rtlanal.c (computed_jump_p): Exit loop once we find label
3686         reference is used.
3688         * i386.c (ix86_pad_returns): Exit loop after setting replace.
3690         * cfgloopmanip.c (remove_path): Exit loop after setting
3691         irred_invalidated.
3693         * gensupport.c (subst_dup): Avoid loop if code is not
3694         MATCH_DUP nor MATCH_OP_DUP.
3696 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
3698         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3700 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3702         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3703         true for SP_REGNUM if mode == ptr_mode.
3704         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3705         with value R0_REGNUM + 31.
3707 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3709         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
3710         pad pointer-typed argument downward.
3712 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3714         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
3715         and __ILP32__ when the ILP32 model is in use.
3717 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3719         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
3720         (aarch64_load_symref_appropriately): In the case of
3721         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
3722         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
3723         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
3724         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
3725         if 'mode' doesn't equal to 'ptr_mode'.
3726         (aarch64_output_mi_thunk): Add an assertion on the alignment of
3727         'vcall_offset'; change to call aarch64_emit_move differently depending
3728         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
3729         to calculate the upper bound of 'vcall_offset'.
3730         (aarch64_cannot_force_const_mem): Change to also return true if
3731         mode != ptr_mode.
3732         (aarch64_legitimize_reload_address): In the case of large
3733         displacements, add new local variable 'xmode' and an assertion
3734         based on it; change to use 'xmode' to generate the new rtx and
3735         reload.
3736         (aarch64_asm_trampoline_template): Change to generate the template
3737         differently depending on TARGET_ILP32 or not; change to use
3738         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
3739         (aarch64_trampoline_size): Removed.
3740         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
3741         and replace immediate literals with it.  Change to use 'ptr_mode'
3742         instead of 'DImode' and call convert_memory_address if the mode
3743         of 'fnaddr' doesn't equal to 'ptr_mode'.
3744         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
3745         to output symbol.
3746         (aarch64_elf_asm_destructor): Likewise.
3747         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
3748         on TARGET_ILP32 instead of aarch64_trampoline_size.
3749         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
3750         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
3751         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
3752         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
3753         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
3754         (storewb_pair<GPI:mode>_<P:mode>): ... this.
3755         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
3756         depending on the value of 'mode'.
3757         (add_losym_<mode>): New.
3758         (ldr_got_small_<mode>): New, based on ldr_got_small.
3759         (ldr_got_small): Remove.
3760         (ldr_got_small_sidi): New.
3761         * config/aarch64/iterators.md (P): New.
3762         (PTR): Change to 'ptr_mode' in the condition.
3764 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3766         * config.gcc (aarch64*-*-*): Support --with-abi.
3767         (aarch64*-*-elf): Support --with-multilib-list.
3768         (aarch64*-*-linux*): Likewise.
3769         (supported_defaults): Add abi to aarch64*-*-*.
3770         * configure.ac: Mention AArch64 for --with-multilib-list.
3771         * configure: Re-generated.
3772         * config/aarch64/biarchilp32.h: New file.
3773         * config/aarch64/biarchlp64.h: New file.
3774         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
3775         (ABI_SPEC): Ditto.
3776         (MULTILIB_DEFAULTS): Ditto.
3777         (DRIVER_SELF_SPECS): Ditto.
3778         (ASM_SPEC): Update to also substitute -mabi.
3779         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
3780         file whose name depends on -mabi= and -mbig-endian.
3781         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
3782         TARGET_ILP32.
3783         (POINTER_SIZE): New define.
3784         (POINTERS_EXTEND_UNSIGNED): Ditto.
3785         (enum aarch64_abi_type): New enumeration tag.
3786         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
3787         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
3788         (TARGET_ILP32): New define.
3789         * config/aarch64/aarch64.opt (mabi): New.
3790         (aarch64_abi): New.
3791         (ilp32, lp64): New values for -mabi.
3792         * config/aarch64/t-aarch64 (comma): New define.
3793         (MULTILIB_OPTIONS): Ditto.
3794         (MULTILIB_DIRNAMES): Ditto.
3795         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
3796         * doc/invoke.texi: Document -mabi for AArch64.
3798 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
3800         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3802 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3803             Anton Blanchard <anton@au1.ibm.com>
3805         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3806         endianness when selecting field to splat.
3808 2013-07-22  Eric Christopher  <echristo@gmail.com>
3810         * dwarf2out.c (die_odr_checksum): New function to use
3811         CHECKSUM_ macros and ULEB128 for DIE tag.
3812         (generate_type_signature): Use.
3814 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
3816         * config.gcc (sparc*-*-*): Accept leon3 processor.
3817         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3818         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3819         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3820         * config/sparc/sparc.opt (enum processor_type): Add leon3.
3821         (mfix-ut699): Adjust comment.
3822         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3823         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3824         (CPP_CPU_SPEC): Likewise.
3825         (ASM_CPU_SPEC): Likewise.
3826         * config/sparc/sparc.c (leon3_cost): New constant.
3827         (sparc_option_override): Add leon3 support.
3828         (mem_ref): New function.
3829         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3830         (sparc_do_work_around_errata): Look into the instruction in the delay
3831         slot and adjust accordingly.  Add fix for the data cache nullify issues
3832         of the UT699.  Change insertion position for the NOP.
3833         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3834         (leon3_load): New reservation.
3835         (leon_store): Bump latency to 2.
3836         (grfpu): New automaton.
3837         (grfpu_alu): New unit.
3838         (grfpu_ds): Likewise.
3839         (leon_fp_alu): Adjust.
3840         (leon_fp_mult): Delete.
3841         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3842         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3843         * config/sparc/sparc.md (cpu): Add leon3.
3844         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3845         (swapsi): Likewise.
3846         (atomic_test_and_set): Likewise.
3847         (ldstub): Likewise.
3849 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
3851         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
3852         default for R5900 targets.
3853         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3854         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3855         * config/mips/mips.c (mips_option_override): Report an error for
3856         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
3857         for -march=r5900 -mhard-float.
3859 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3861         * df-problems.c (can_move_insns_across): Exit loop once we
3862         find a non-fixed, non-global register.
3864         * ipa-pure-const.c (propagate_nothrow): Exit loop after
3865         setting can_throw.
3867         * omega.c (omega_eliminate_red): Break after setting red_found.
3868         (omega_problem_has_red_equations): Similarly after setting found.
3869         (omega_query_variable): Similarly after setting coupled.
3871 2013-07-22  Marek Polacek  <polacek@redhat.com>
3873         * gimplify.c: Don't include gimple.h twice.
3875 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3877         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3878         instead of TARGET_THUMB1.
3879         (Pz): New constraint.
3880         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3881         encodings.
3882         (compare_negsi_si): Likewise.
3883         (compare_addsi2_op0): Likewise.
3884         (compare_addsi2_op1): Likewise.
3885         (addsi3_carryin_<optab>): Likewise.
3886         (addsi3_carryin_alt2_<optab>): Likewise.
3887         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3888         for arm_restrict_it.
3889         (subsi3_carryin): Likewise.
3890         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3891         (minmax_arithsi): Disable for arm_restrict_it.
3892         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3893         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3894         (satsi_<SAT:code>_shift): Likewise.
3895         (arm_shiftsi3): Add alternative for 16-bit encoding.
3896         (arm32_movhf): Disable for arm_restrict_it.
3897         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3898         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3900 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
3902         * config/arm/arm.md (attribute "insn"): Delete.
3903         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3904         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3905         (not_shiftsi): Update for attribute change.
3906         (not_shiftsi_compare0): Likewise.
3907         (not_shiftsi_compare0_scratch): Likewise.
3908         (arm_one_cmplsi2): Likewise.
3909         (thumb1_one_cmplsi2): Likewise.
3910         (notsi_compare0): Likewise.
3911         (notsi_compare0_scratch): Likewise.
3912         (thumb1_movdi_insn): Likewise.
3913         (arm_movsi_insn): Likewise.
3914         (movhi_insn_arch4): Likewise.
3915         (movhi_bytes): Likewise.
3916         (arm_movqi_insn): Likewise.
3917         (thumb1_movqi_insn): Likewise.
3918         (arm32_movhf): Likewise.
3919         (thumb1_movhf): Likewise.
3920         (arm_movsf_soft_insn): Likewise.
3921         (thumb1_movsf_insn): Likewise.
3922         (thumb_movdf_insn): Likewise.
3923         (movsicc_insn): Likewise.
3924         (movsfcc_soft_insn): Likewise.
3925         (and_scc): Likewise.
3926         (cond_move): Likewise.
3927         (if_move_not): Likewise.
3928         (if_not_move): Likewise.
3929         (if_shift_move): Likewise.
3930         (if_move_shift): Likewise.
3931         (if_shift_shift): Likewise.
3932         (if_not_arith): Likewise.
3933         (if_arith_not): Likewise.
3934         (cond_move_not): Likewise.
3935         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3936         (neon_mov<mode>): Likewise.
3937         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3938         (thumb2_movsi_vfp): Likewise.
3939         (movsf_vfp): Likewise.
3940         (thumb2_movsf_vfp): Likewise.
3941         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3942         change.
3943         (cortexa7_older_only): Likewise.
3944         (cortexa7_younger): Likewise.
3945         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3946         (1020alu_shift_op): Likewise.
3947         (1020alu_shift_reg_op): Likewise.
3948         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3949         (alu_shift_op): Likewise.
3950         (alu_shift_reg_op): Likewise.
3951         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3952         (11_alu_shift_op): Likewise.
3953         (11_alu_shift_reg_op): Likewise.
3954         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3955         (9_alu_shift_reg_op): Likewise.
3956         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3957         change.
3958         (cortex_a15_alu_shift): Likewise.
3959         (cortex_a15_alu_shift_reg): Likewise.
3960         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3961         (cortex_a5_alu_shift): Likewise.
3962         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3963         change.
3964         (cortex_a53_alu_shift): Likewise.
3965         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3966         change.
3967         (cortex_a7_alu_reg): Likewise.
3968         (cortex_a7_alu_shift): Likewise.
3969         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3970         (cortex_a8_alu_shift): Likewise.
3971         (cortex_a8_alu_shift_reg): Likewise.
3972         (cortex_a8_mov): Likewise.
3973         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3974         (cortex_a9_dp_shift): Likewise.
3975         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3976         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3977         (cortex_r4_mov): Likewise.
3978         (cortex_r4_alu_shift): Likewise.
3979         (cortex_r4_alu_shift_reg): Likewise.
3980         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3981         (526_alu_shift_op): Likewise.
3982         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3983         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3984         (626te_alu_shift_op): Likewise.
3985         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3986         (726te_alu_op): Likewise.
3987         (726te_alu_shift_op): Likewise.
3988         (726te_alu_shift_reg_op): Likewise.
3989         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3990         (mp626_alu_shift_op): Likewise.
3991         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3992         (pj4_alu_e1_conds): Likewise.
3993         (pj4_alu): Likewise.
3994         (pj4_alu_conds): Likewise.
3995         (pj4_shift): Likewise.
3996         (pj4_shift_conds): Likewise.
3997         (pj4_alu_shift): Likewise.
3998         (pj4_alu_shift_conds): Likewise.
4000 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4002         * config/arm/predicates.md (shiftable_operator_strict_it):
4003         New predicate.
4004         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4005         Disable cond_exec version for arm_restrict_it.
4006         (thumb2_smaxsi3): Convert to generate cond_exec.
4007         (thumb2_sminsi3): Likewise.
4008         (thumb32_umaxsi3): Likewise.
4009         (thumb2_uminsi3): Likewise.
4010         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4011         (thumb2_neg_abssi2): Likewise.
4012         (thumb2_mov_scc): Add alternative for 16-bit encoding.
4013         (thumb2_movsicc_insn): Adjust alternatives.
4014         (thumb2_mov_negscc): Disable for arm_restrict_it.
4015         (thumb2_mov_negscc_strict_it): New pattern.
4016         (thumb2_mov_notscc_strict_it): New pattern.
4017         (thumb2_mov_notscc): Disable for arm_restrict_it.
4018         (thumb2_ior_scc): Likewise.
4019         (thumb2_ior_scc_strict_it): New pattern.
4020         (thumb2_cond_move): Adjust for arm_restrict_it.
4021         (thumb2_cond_arith): Disable for arm_restrict_it.
4022         (thumb2_cond_arith_strict_it): New pattern.
4023         (thumb2_cond_sub): Adjust for arm_restrict_it.
4024         (thumb2_movcond): Likewise.
4025         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4026         (thumb2_zero_extendhisi2_v6): Likewise.
4027         (thumb2_zero_extendqisi2_v6): Likewise.
4028         (orsi_notsi_si): Likewise.
4029         (orsi_not_shiftsi_si): Likewise.
4031 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
4033         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4034         instruction sequence is 1 byte shorter.
4036 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
4038         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4039         it is not needed after split.
4041 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
4043         PR target/51784
4044         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4045         second label for nonlocal goto receivers. Don't output pic base labels
4046         unless we're producing PIC; mark that action unreachable().
4047         (ix86_save_reg): If the function contains a nonlocal label, save the
4048         PIC base reg.
4049         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4050         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4051         (update_pic_label_number_if_needed): New.
4052         (machopic_output_function_base_name): Adjust for nonlocal receiver
4053         case.
4054         (machopic_should_output_picbase_label): New.
4055         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4056         (nonlocal_goto_receiver): New insn and split.
4058 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
4060         * config/aarch64/aarch64-builtins.c
4061         (aarch64_fold_builtin): Fold abs in all modes.
4062         * config/aarch64/aarch64-simd-builtins.def
4063         (abs): Enable for all modes.
4064         * config/aarch64/arm_neon.h
4065         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4066         (vabs_f64): Add missing intrinsic.
4068 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4070         * config/aarch64/arm_neon.h (vabs_s64): New function
4072 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4074         PR target/57516
4075         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4076         * config/avr/avr.md (adjust_len): Add `round'.
4077         * config/avr/avr-protos.h (avr_out_round): New prototype.
4078         (avr_out_plus): Add `out_label' argument.
4079         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4080         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4081         Handle the case where `insn' is just a pattern.
4082         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4083         (avr_out_round): New function.
4084         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4086 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
4088         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4089         Add check for flag_stack_usage to handle -fstack-usage support
4091 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
4093         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4094         interaction for new Power8 flags and VSX.
4096 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4098         PR middle-end/57698
4099         * tree-inline.c (expand_call_inline): Emit errors during
4100         early_inlining only if optimization is not turned on.
4102 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4104         * passes.def: New.
4106         * passes.c (init_optimization_passes): Move the construction of
4107         the pass hierarchy into a new passes.def file.
4109         * Makefile.in (passes.o): Add dependency on passes.def.
4111 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4113         * passes.c (init_optimization_passes): Introduce macros for
4114         constructing the tree of passes (INSERT_PASSES_AFTER,
4115         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4116         TERMINATE_PASS_LIST).
4118 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
4119             Wei Mi  <wmi@google.com>
4121         PR rtl-optimization/57878
4122         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4123         top.
4124         (reload_pseudo_compare_func): Check nregs first for reload
4125         pseudos.
4127 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4129         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4131 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
4133         * read-rtl.c (validate_const_int): Once an invalid character is
4134         seen, quit the loop.
4136         * gengtype.c (write_roots): Similarly once we find the "deletable"
4137         or "if_marked" option.
4139 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4141         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4142         "xtab" and "sat".  Move value "clz" from here to ...
4143         (attriubte "type"): ... here.
4144         (satsi_<SAT:code>): Delete "insn" attribute.
4145         (satsi_<SAT:code>_shift): Likewise.
4146         (arm_zero_extendqisi2addsi): Likewise.
4147         (arm_extendqisi2addsi): Likewise.
4148         (clzsi2): Update for attribute changes.
4149         (rbitsi2): Likewise.
4150         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4151         attribute.
4152         (arm_usatsihi): Likewise.
4153         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4155 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4157         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4158         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
4159         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
4160         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
4161         in alphabetical order.
4162         (attribute "core_cycles"): Update for attribute changes.
4163         (arm_addsi3): Likewise.
4164         (addsi3_compare0): Likewise.
4165         (addsi3_compare0_scratch): Likewise.
4166         (addsi3_compare_op1): Likewise.
4167         (addsi3_compare_op2): Likewise.
4168         (compare_addsi2_op0): Likewise.
4169         (compare_addsi2_op1): Likewise.
4170         (addsi3_carryin_shift_<optab>): Likewise.
4171         (subsi3_carryin_shift): Likewise.
4172         (rsbsi3_carryin_shift): Likewise.
4173         (arm_subsi3_insn): Likewise.
4174         (subsi3_compare0): Likewise.
4175         (subsi3_compare): Likewise.
4176         (arm_andsi3_insn): Likewise.
4177         (thumb1_andsi3_insn): Likewise.
4178         (andsi3_compare0): Likewise.
4179         (andsi3_compare0_scratch): Likewise.
4180         (zeroextractsi_compare0_scratch
4181         (andsi_not_shiftsi_si): Likewise.
4182         (iorsi3_insn): Likewise.
4183         (iorsi3_compare0): Likewise.
4184         (iorsi3_compare0_scratch): Likewise.
4185         (arm_xorsi3): Likewise.
4186         (thumb1_xorsi3_insn): Likewise.
4187         (xorsi3_compare0): Likewise.
4188         (xorsi3_compare0_scratch): Likewise.
4189         (satsi_<SAT:code>_shift): Likewise.
4190         (rrx): Likewise.
4191         (arm_shiftsi3): Likewise.
4192         (shiftsi3_compare0): Likewise.
4193         (not_shiftsi): Likewise.
4194         (not_shiftsi_compare0): Likewise.
4195         (not_shiftsi_compare0_scratch): Likewise.
4196         (arm_one_cmplsi2): Likewise.
4197         (thumb_one_complsi2): Likewise.
4198         (notsi_compare0): Likewise.
4199         (notsi_compare0_scratch): Likewise.
4200         (thumb1_zero_extendhisi2): Likewise.
4201         (arm_zero_extendhisi2): Likewise.
4202         (arm_zero_extendhisi2_v6): Likewise.
4203         (arm_zero_extendhisi2addsi): Likewise.
4204         (thumb1_zero_extendqisi2): Likewise.
4205         (thumb1_zero_extendqisi2_v6): Likewise.
4206         (arm_zero_extendqisi2): Likewise.
4207         (arm_zero_extendqisi2_v6): Likewise.
4208         (arm_zero_extendqisi2addsi): Likewise.
4209         (thumb1_extendhisi2): Likewise.
4210         (arm_extendhisi2): Likewise.
4211         (arm_extendhisi2_v6): Likewise.
4212         (arm_extendqisi): Likewise.
4213         (arm_extendqisi_v6): Likewise.
4214         (arm_extendqisi2addsi): Likewise.
4215         (thumb1_extendqisi2): Likewise.
4216         (thumb1_movdi_insn): Likewise.
4217         (arm_movsi_insn): Likewise.
4218         (movsi_compare0): Likewise.
4219         (movhi_insn_arch4): Likewise.
4220         (movhi_bytes): Likewise.
4221         (arm_movqi_insn): Likewise.
4222         (thumb1_movqi_insn): Likewise.
4223         (arm32_movhf): Likewise.
4224         (thumb1_movhf): Likewise.
4225         (arm_movsf_soft_insn): Likewise.
4226         (thumb1_movsf_insn): Likewise.
4227         (movdf_soft_insn): Likewise.
4228         (thumb_movdf_insn): Likewise.
4229         (arm_cmpsi_insn): Likewise.
4230         (cmpsi_shiftsi): Likewise.
4231         (cmpsi_shiftsi_swp): Likewise.
4232         (arm_cmpsi_negshiftsi_si): Likewise.
4233         (movsicc_insn): Likewise.
4234         (movsfcc_soft_insn): Likewise.
4235         (arith_shiftsi): Likewise.
4236         (arith_shiftsi_compare0
4237         (arith_shiftsi_compare0_scratch
4238         (sub_shiftsi): Likewise.
4239         (sub_shiftsi_compare0
4240         (sub_shiftsi_compare0_scratch
4241         (and_scc): Likewise.
4242         (cond_move): Likewise.
4243         (if_plus_move): Likewise.
4244         (if_move_plus): Likewise.
4245         (if_move_not): Likewise.
4246         (if_not_move): Likewise.
4247         (if_shift_move): Likewise.
4248         (if_move_shift): Likewise.
4249         (if_shift_shift): Likewise.
4250         (if_not_arith): Likewise.
4251         (if_arith_not): Likewise.
4252         (cond_move_not): Likewise.
4253         (thumb1_ashlsi3): Set type attribute.
4254         (thumb1_ashrsi3): Likewise.
4255         (thumb1_lshrsi3): Likewise.
4256         (thumb1_rotrsi3): Likewise.
4257         (shiftsi3_compare0_scratch): Likewise.
4258         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4259         (neon_mov<mode>): Likewise.
4260         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4261         attribute changes.
4262         (thumb2_movsi_insn): Likewise.
4263         (thumb2_cmpsi_neg_shiftsi): Likewise.
4264         (thumb2_extendqisi_v6): Likewise.
4265         (thumb2_zero_extendhisi2_v6): Likewise.
4266         (thumb2_zero_extendqisi2_v6): Likewise.
4267         (thumb2_shiftsi3_short): Likewise.
4268         (thumb2_addsi3_compare0_scratch): Likewise.
4269         (orsi_not_shiftsi_si): Likewise.
4270         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4271         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4272         changes.
4273         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4274         (1020alu_shift_op): Likewise.
4275         (1020alu_shift_reg_op): Likewise.
4276         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4277         (alu_shift_op): Likewise.
4278         (alu_shift_reg_op): Likewise.
4279         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4280         (11_alu_shift_op): Likewise.
4281         (11_alu_shift_reg_op): Likewise.
4282         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4283         (9_alu_shift_reg_op): Likewise.
4284         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4285         changes.
4286         (cortex_a15_alu_shift): Likewise.
4287         (cortex_a15_alu_shift_reg): Likewise.
4288         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4289         changes.
4290         (cortex_a5_alu_shift): Likewise.
4291         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4292         changes.
4293         (cortex_a53_alu_shift): Likewise.
4294         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4295         changes.
4296         (cortex_a7_alu_reg): Likewise.
4297         (cortex_a7_alu_shift): Likewise.
4298         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4299         changes.
4300         (cortex_a8_alu_shift): Likewise.
4301         (cortex_a8_alu_shift_reg): Likewise.
4302         (cortex_a8_mov): Likewise.
4303         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
4304         (cortex_a9_dp_shift): Likewise.
4305         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
4306         changes.
4307         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
4308         changes.
4309         (cortex_r4_mov): Likewise.
4310         (cortex_r4_alu_shift): Likewise.
4311         (cortex_r4_alu_shift_reg): Likewise.
4312         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
4313         (526_alu_shift_op): Likewise.
4314         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
4315         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
4316         (626te_alu_shift_op): Likewise.
4317         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
4318         (726te_alu_op): Likewise.
4319         (726te_alu_shift_op): Likewise.
4320         (726te_alu_shift_reg_op): Likewise.
4321         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
4322         (mp626_alu_shift_op): Likewise.
4323         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
4324         (pj4_alu_e1_conds): Likewise.
4325         (pj4_alu): Likewise.
4326         (pj4_alu_conds): Likewise.
4327         (pj4_shift): Likewise.
4328         (pj4_shift_conds): Likewise.
4329         (pj4_alu_shift): Likewise.
4330         (pj4_alu_shift_conds): Likewise.
4331         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4332         changes.
4333         (cortexa7_older_only): Likewise.
4334         (cortexa7_younger): Likewise.
4336 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4338         * ipa-pure-const.c (generate_summary): Rename to...
4339         (pure_const_generate_summary): ... this.
4341 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
4343         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
4345 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
4347         PR target/57909
4348         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
4349         usage in HI mode.
4351 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4353         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
4354         enabled without -march=zEC12.
4355         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
4356         flags to be set.
4358 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4360         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
4361         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
4362         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
4363         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
4364         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
4365         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
4366         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
4367         ISA_HAS_FP4.
4368         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
4369         and ISA_HAS_NMADD3_NMSUB3.
4370         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
4371         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
4372         (nmsub4<mode>, nmsub3<mode>): Likewise.
4373         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
4375 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4377         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
4378         TARGET_MIPS5400 checking.
4380 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
4381             Peter Bergner  <bergner@vnet.ibm.com>
4383         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
4384         registers in the comment.
4385         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
4386         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
4387         rather than FIRST_PSEUDO_REGISTERS.
4389 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
4391         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
4392         enable extra ISA flags with TARGET_HTM.
4394 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4396         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
4397         Fix comment typos.
4399 2013-07-15  Cong Hou  <congh@google.com>
4401         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
4402         in compare function for sorting.
4404 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
4406         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
4407         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
4408         * config/rs6000/rs6000.opt: Add -mhtm option.
4409         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
4410         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
4411         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
4412         __HTM__ if the HTM instructions are available.
4413         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
4414         htm_spr_reg_operand): New define_predicates.
4415         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
4416         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
4417         Include htm.md.
4418         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
4419         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
4420         HTM builtin functions.
4421         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
4422         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
4423         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
4424         (rs6000_builtin_mask_calculate): Likewise.
4425         (rs6000_option_override_internal): Likewise.
4426         (bdesc_htm): Add new HTM builtin support.
4427         (htm_spr_num): New function.
4428         (htm_spr_regno): Likewise.
4429         (rs6000_htm_spr_icode): Likewise.
4430         (htm_expand_builtin): Likewise.
4431         (htm_init_builtins): Likewise.
4432         (rs6000_expand_builtin): Add support for HTM builtin functions.
4433         (rs6000_init_builtins): Likewise.
4434         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
4435         option.
4436         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
4437         (TARGET_HTM, MASK_HTM): Define macros.
4438         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
4439         (FIXED_REGISTERS): Likewise.
4440         (CALL_USED_REGISTERS): Likewise.
4441         (CALL_REALLY_USED_REGISTERS): Likewise.
4442         (REG_ALLOC_ORDER): Likewise.
4443         (enum reg_class): Likewise.
4444         (REG_CLASS_NAMES): Likewise.
4445         (REG_CLASS_CONTENTS): Likewise.
4446         (REGISTER_NAMES): Likewise.
4447         (ADDITIONAL_REGISTER_NAMES): Likewise.
4448         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
4449         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
4450         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
4451         * config/rs6000/htm.md: New file.
4452         * config/rs6000/htmintrin.h: New file.
4453         * config/rs6000/htmxlintrin.h: New file.
4455 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4457         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4458         Define SYMBOL_TINY_GOT, update comment.
4459         * config/aarch64/aarch64.c
4460         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
4461         (aarch64_expand_mov_immediate): Likewise.
4462         (aarch64_print_operand): Likewise.
4463         (aarch64_classify_symbol): Likewise.
4464         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
4465         (ldr_got_tiny): Define.
4467 2013-07-13  Tobias Grosser  <tobias@grosser.es>
4469         PR tree-optimization/54094
4470         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
4471           scheduling dimension for the parallelism check from the polyhedral
4472           information in the AST.
4473         * graphite-dependences.c (carries_deps): Do not assume the schedule is
4474           in 2D + 1 form.
4476 2013-07-13  Jason Merrill  <jason@redhat.com>
4478         * print-tree.c (debug_vec_tree): Use debug_raw.
4479         (debug_raw (vec<tree, va_gc> &)): New.
4480         (debug_raw (vec<tree, va_gc> *)): New.
4481         * tree.h: Declare them.
4483 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
4485         * ifcvt.c (ifcvt_after_combine): New static variable.
4486         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
4487         for size.
4488         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
4489         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
4490         rest_of_handle_if_after_reload): Pass new argument for if_convert.
4492 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
4494         * config/mips/mips.c (mips_expand_call): Remove empty statement.
4496 2013-07-12  Michael Matz  <matz@suse.de>
4498         PR middle-end/55771
4499         * convert.c (convert_to_real): Reject non-float inner types.
4501 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
4503         * config/aarch64/aarch64-protos.h
4504         (aarch64_simd_immediate_valid_for_move): Remove.
4505         * config/aarch64/aarch64.c (simd_immediate_info): New member.
4506         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
4507         cases.
4508         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
4510 2013-07-11  Steve Ellcey  <sellcey@mips.com>
4512         * config/mips/mips.c (mips_conditional_register_usage): Do not
4513         use t[0-7] registers in MIPS16 mode when optimizing for size.
4515 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
4517         * config/i386/i386.c (dispatch_function_versions): Fix array
4518         indexing of function_version_info to match actual_versions.
4520 2013-07-11  Teresa Johnson  <tejohnson@google.com>
4522         * vec.h (struct va_gc): Move release out-of-line.
4523         (va_gc::release): Call ggc_free on released vec.
4525 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4527         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
4528         Require GOT register as additional operand in UNSPEC.
4529         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
4530         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4531         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4532         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
4533         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
4534         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
4535         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
4537 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
4539         PR target/57631
4540         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
4541         name seen by assembler/linker rather if available.
4543 2013-07-11  Andreas Schwab  <schwab@suse.de>
4545         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4547 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
4549         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4551 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
4553         * doc/tm.texi.in: Move hook documentation to ....
4554         * target.def: ... here.
4556         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4557         text on @hook line.
4558         * doc/tm.texi: Regenerate.
4560 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
4562         PR c++/57869
4563         * doc/invoke.texi: Document -Wconditionally-supported.
4565 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4567         PR target/57844
4568         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4569         of my_fp.
4571 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4573         PR target/57506
4574         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4575         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4576         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4577         Remove duplicate devices.
4578         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4579         * config/avr/t-multilib: Regenerate.
4580         * config/avr/avr-tables.opt: Regenerate.
4581         * doc/avr-mmcu.texi: Regenerate.
4583 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4585         PR target/56987
4586         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4588 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
4590         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4591         the cost of MULT when optimizing for size.
4593 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4595         * config/cr16/cr16-protos.h: Don't include target.h.
4597 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
4599         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4600         adjust register size for TDmode and TFmode for VSX registers.
4602 2013-07-08  Kai Tietz  <ktietz@redhat.com>
4604         PR target/56892
4605         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4606         hook_bool_const_tree_true.
4608 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4610         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4611         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4612         * config/s390/s390.md: Define FPR*_REGNUM constants.
4613         Fix FPR2_REGNUM constant (18 -> 17).
4614         ("*trunc<BFP:mode><DFP_ALL:mode>2")
4615         ("*trunc<DFP_ALL:mode><BFP:mode>2")
4616         ("trunc<BFP:mode><DFP_ALL:mode>2")
4617         ("trunc<DFP_ALL:mode><BFP:mode>2")
4618         ("*extend<BFP:mode><DFP_ALL:mode>2")
4619         ("*extend<DFP_ALL:mode><BFP:mode>2")
4620         ("extend<BFP:mode><DFP_ALL:mode>2")
4621         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4622         FPR4_REGNUM.
4624 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
4626         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4628 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4630         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4631         and cfun_fpr_bit_p to cfun_fpr_save_p.
4632         (s390_frame_area, s390_register_info, s390_frame_info)
4633         (s390_emit_prologue, s390_emit_epilogue)
4634         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4635         register numbers.
4636         * config/s390/s390.h: Define *_REGNUM macros for floating point
4637         register numbers.
4639 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
4641         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4643 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
4645         PR rtl-optimization/57786
4646         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4647         and break out of the loop when it is set to false.
4649 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
4651         PR target/57819
4652         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4653         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4654         (const_int 63)) 0)).
4655         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4656         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4657         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4659         PR rtl-optimization/57829
4660         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4661         mask bits outside of mode are just sign-extension from mode to HWI.
4663 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
4665         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4666         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4667         adjust_address instead of change_address to keep info about alignment.
4668         (emit_strmov): Remove.
4669         (emit_memmov): New function.
4670         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4671         (expand_movmem_epilogue): Likewise and return updated rtx for
4672         destination.
4673         (expand_constant_movmem_prologue): Likewise and return updated rtx for
4674         destination and source.
4675         (decide_alignment): Refactor, handle vector_loop.
4676         (ix86_expand_movmem): Likewise.
4677         (ix86_expand_setmem): Likewise.
4678         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4680 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
4682         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4683         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4685 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4687         * config/i386/sse.md (sse_movlhps): Change alternative 3
4688         of operand 2 to "m".
4690 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4692         PR target/57807
4693         * config/i386/sse.md (iptr): New mode attribute.
4694         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4695         (<sse>_vm<plusminus_insn><mode>3): Ditto.
4696         (<sse>_vmmul<mode>3): Ditto.
4697         (<sse>_vmdiv<mode>3): Ditto.
4698         (sse_vmrcpv4sf2): Ditto.
4699         (<sse>_vmsqrt<mode>2): Ditto.
4700         (sse_vmrsqrtv4sf2): Ditto.
4701         (<sse>_vm<code><mode>3): Ditto.
4702         (avx_vmcmp<mode>3): Ditto.
4703         (<sse>_vmmaskcmp<mode>3): Ditto.
4704         (<sse>_comi): Ditto.
4705         (<sse>_ucomi): Ditto.
4706         (*xop_vmfrcz_<mode>): Ditto.
4707         (*fmai_fmadd_<mode>): Ditto.
4708         (*fmai_fmsub_<mode>): Ditto.
4709         (*fmai_fnmadd_<mode>): Ditto.
4710         (*fmai_fnmsub_<mode>): Ditto.
4711         (*fma4i_vmfmadd_<mode>): Ditto.
4712         (*fma4i_vmfmsub_<mode>): Ditto.
4713         (*fma4i_vmfnmadd_<mode>): Ditto.
4714         (*fma4i_vmfnmsub_<mode>): Ditto.
4715         (*xop_vmfrcz_<mode>): Ditto.
4716         (sse_cvtps2pi): Ditto.
4717         (sse_cvttps2pi): Ditto.
4718         (sse_cvtss2si): Ditto.
4719         (sse_cvtss2si_2): Ditto.
4720         (sse_cvtss2siq_2): Ditto.
4721         (sse_cvttss2si): Ditto.
4722         (sse_cvttss2siq): Ditto.
4723         (sse_cvtsd2si): Ditto.
4724         (sse_cvtsd2si_2): Ditto.
4725         (sse_cvtsd2siq_2): Ditto.
4726         (sse_cvttsd2si): Ditto.
4727         (sse_cvttsd2siq): Ditto.
4728         (sse_cvtsd2ss): Ditto.
4729         (sse_cvtss2sd): Ditto.
4730         (avx2_pbroadcast<mode>): Ditto.
4731         (avx2_pbroadcast<mode>_1): Ditto.
4732         (*avx_vperm_broadcast_v4sf): Ditto.
4734         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
4735         (sse_movlhps): Ditto.
4736         (sse_storehps): Ditto.
4737         (sse_loadhps): Ditto.
4738         (sse_storelps): Ditto.
4739         (sse_loadlps): Ditto.
4740         (*vec_concatv4sf): Ditto.
4741         (*vec_interleave_highv2df): Ditto.
4742         (*vec_interleave_lowv2df): Ditto.
4743         (*vec_extractv2df_1_sse): Ditto.
4744         (*vec_extractv2df_0_sse): Ditto.
4745         (sse2_storelpd): Ditto.
4746         (sse2_loadlpd): Ditto.
4747         (sse2_movsd): Ditto.
4748         (*vec_concatv4si): Ditto.
4749         (vec_concatv2di): Ditto.
4751         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
4752         for Intel asm dialect.
4753         (mmx_punpcklwd): Ditto.
4754         (mmx_punpckldq): Ditto.
4756         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
4757         for intel assembler dialect.
4759 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
4761         PR target/29776
4762         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
4763         for BUILT_IN_C{LZ,LRSB}*.
4764         * tree.h (CASE_INT_FN): Add FN##IMAX case.
4765         * tree-vrp.c (extract_range_basic): Handle
4766         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
4767         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
4768         fall thru to code calling set_value*.
4769         * builtins.c (expand_builtin): Remove *IMAX cases.
4770         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
4771         if width is bigger than 2*HWI.
4773 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
4775         PR rtl-optimization/55342
4776         * lra-int.h (lra_subreg_reload_pseudos): New.
4777         * lra.c: Add undoing optional reloads to the block diagram.
4778         (lra_subreg_reload_pseudos): New.
4779         (lra_optional_reload_pseudos): Change comments.
4780         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
4781         lra_optional_reload_pseudos after undo transformations.
4782         * lra-assigns.c (pseudo_prefix_title): New.
4783         (lra_setup_reg_renumber): Use it.
4784         (spill_for): Ditto.  Check subreg reload pseudos too.
4785         (assign_by_spills): Consider subreg reload pseudos too.
4786         * lra-constraints.c (simplify_operand_subreg): Use
4787         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
4788         (curr_insn_transform): Recognize and do optional reloads.
4789         (undo_optional_reloads): New.
4790         (lra_undo_inheritance): Call undo_optional_reloads.
4792 2013-07-05  Thomas Quinot  <quinot@adacore.com>
4794         * tree-complex.c (expand_complex_operations_1): Fix typo.
4796 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
4798         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
4799         (tune_params): New member 'const vec_costs'.
4800         * config/aarch64/aarch64.c (generic_vector_cost): New.
4801         (generic_tunings): New member 'generic_vector_cost'.
4802         (aarch64_builtin_vectorization_cost): New.
4803         (aarch64_add_stmt_cost): New.
4804         (TARGET_VECTORIZE_ADD_STMT_COST): New.
4805         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4807 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
4809         PR target/57777
4810         * config/i386/predicates.md (vsib_address_operand): Disallow
4811         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4813 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
4815         PR middle-end/55030
4816         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4817         expand_builtin_setjmp_receiver.
4818         (expand_label): Adjust, call expand_builtin_setjmp_receiver
4819         with NULL for the label parameter.
4820         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4821         the frame-pointer.  Adjust comments.
4822         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4823         only if LABEL is non-NULL.
4825 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
4827         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4828         (ARM_ABI_AAPCS64): Ditto.
4829         (arm_abi): Ditto.
4830         (ARM_DEFAULT_ABI): Ditto.
4832 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
4834         * config/aarch64/aarch64-builtins.c
4835         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4836         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4837         (st1): Likewise.
4838         * config/aarch64/aarch64-simd.md
4839         (aarch64_ld1<VALL:mode>): New.
4840         (aarch64_st1<VALL:mode>): Likewise.
4841         * config/aarch64/arm_neon.h
4842         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4844 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
4846         * config/i386/i386.c (gate_insert_vzeroupper): Check if
4847         target ISA is AVX.
4848         (ix86_option_override_internal):Turn on all -mavx target flags by
4849         default as they are dependent on AVX anyway.
4851 2013-07-02  Cary Coutant  <ccoutant@google.com>
4853         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4854         deterministic hash.
4855         (loc_checksum_ordered): Likewise.
4856         (hash_loc_operands): Remove inline keyword.
4858 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
4860         PR tree-optimization/57741
4861         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4862         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4863         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4864         Allow REAL_CST step_exprs if flag_associative_math.
4865         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4867 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4869         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4871 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4873         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4875 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4877         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4878         encoding.
4879         (iorsi3_insn): Likewise.
4880         (arm_xorsi3): Likewise.
4882 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
4884         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
4885         to ...
4886         (attribute "type"): ... here, and prefix with "wmmx_".
4887         (attribute "core_cycles"): Update for attribute changes.
4888         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4889         (tbcstv4hi): Likewise.
4890         (tbcstv2si): Likewise.
4891         (iwmmxt_iordi3): Likewise.
4892         (iwmmxt_xordi3): Likewise.
4893         (iwmmxt_anddi3): Likewise.
4894         (iwmmxt_nanddi3): Likewise.
4895         (iwmmxt_arm_movdi): Likewise.
4896         (iwmmxt_movsi_insn): Likewise.
4897         (mov<mode>_internal): Likewise.
4898         (and<mode>3_iwmmxt): Likewise.
4899         (ior<mode>3_iwmmxt): Likewise.
4900         (xor<mode>3_iwmmxt): Likewise.
4901         (add<mode>3_iwmmxt): Likewise.
4902         (ssaddv8qi3): Likewise.
4903         (ssaddv4hi3): Likewise.
4904         (ssaddv2si3): Likewise.
4905         (usaddv8qi3): Likewise.
4906         (usaddv4hi3): Likewise.
4907         (usaddv2si3): Likewise.
4908         (sub<mode>3_iwmmxt): Likewise.
4909         (sssubv8qi3): Likewise.
4910         (sssubv4hi3): Likewise.
4911         (sssubv2si3): Likewise.
4912         (ussubv8qi3): Likewise.
4913         (ussubv4hi3): Likewise.
4914         (ussubv2si3): Likewise.
4915         (mulv4hi3_iwmmxt): Likewise.
4916         (smulv4hi3_highpart): Likewise.
4917         (umulv4hi3_highpart): Likewise.
4918         (iwmmxt_wmacs): Likewise.
4919         (iwmmxt_wmacsz): Likewise.
4920         (iwmmxt_wmacu): Likewise.
4921         (iwmmxt_wmacuz): Likewise.
4922         (iwmmxt_clrdi): Likewise.
4923         (iwmmxt_clrv8qi): Likewise.
4924         (iwmmxt_clr4hi): Likewise.
4925         (iwmmxt_clr2si): Likewise.
4926         (iwmmxt_uavgrndv8qi3): Likewise.
4927         (iwmmxt_uavgrndv4hi3): Likewise.
4928         (iwmmxt_uavgv8qi3): Likewise.
4929         (iwmmxt_uavgv4hi3): Likewise.
4930         (iwmmxt_tinsrb): Likewise.
4931         (iwmmxt_tinsrh): Likewise.
4932         (iwmmxt_tinsrw): Likewise.
4933         (iwmmxt_textrmub): Likewise.
4934         (iwmmxt_textrmsb): Likewise.
4935         (iwmmxt_textrmuh): Likewise.
4936         (iwmmxt_textrmsh): Likewise.
4937         (iwmmxt_textrmw): Likewise.
4938         (iwmxxt_wshufh): Likewise.
4939         (eqv8qi3): Likewise.
4940         (eqv4hi3): Likewise.
4941         (eqv2si3): Likewise.
4942         (gtuv8qi3): Likewise.
4943         (gtuv4hi3): Likewise.
4944         (gtuv2si3): Likewise.
4945         (gtv8qi3): Likewise.
4946         (gtv4hi3): Likewise.
4947         (gtv2si3): Likewise.
4948         (smax<mode>3_iwmmxt): Likewise.
4949         (umax<mode>3_iwmmxt): Likewise.
4950         (smin<mode>3_iwmmxt): Likewise.
4951         (umin<mode>3_iwmmxt): Likewise.
4952         (iwmmxt_wpackhss): Likewise.
4953         (iwmmxt_wpackwss): Likewise.
4954         (iwmmxt_wpackdss): Likewise.
4955         (iwmmxt_wpackhus): Likewise.
4956         (iwmmxt_wpackwus): Likewise.
4957         (iwmmxt_wpackdus): Likewise.
4958         (iwmmxt_wunpckihb): Likewise.
4959         (iwmmxt_wunpckihh): Likewise.
4960         (iwmmxt_wunpckihw): Likewise.
4961         (iwmmxt_wunpckilb): Likewise.
4962         (iwmmxt_wunpckilh): Likewise.
4963         (iwmmxt_wunpckilw): Likewise.
4964         (iwmmxt_wunpckehub): Likewise.
4965         (iwmmxt_wunpckehuh): Likewise.
4966         (iwmmxt_wunpckehuw): Likewise.
4967         (iwmmxt_wunpckehsb): Likewise.
4968         (iwmmxt_wunpckehsh): Likewise.
4969         (iwmmxt_wunpckehsw): Likewise.
4970         (iwmmxt_wunpckelub): Likewise.
4971         (iwmmxt_wunpckeluh): Likewise.
4972         (iwmmxt_wunpckeluw): Likewise.
4973         (iwmmxt_wunpckelsb): Likewise.
4974         (iwmmxt_wunpckelsh): Likewise.
4975         (iwmmxt_wunpckelsw): Likewise.
4976         (ror<mode>3): Likewise.
4977         (ashr<mode>3_iwmmxt): Likewise.
4978         (lshr<mode>3_iwmmxt): Likewise.
4979         (ashl<mode>3_iwmmxt): Likewise.
4980         (ror<mode>3_di): Likewise.
4981         (ashr<mode>3_di): Likewise.
4982         (lshr<mode>3_di): Likewise.
4983         (ashl<mode>3_di): Likewise.
4984         (iwmmxt_wmadds): Likewise.
4985         (iwmmxt_wmaddu): Likewise.
4986         (iwmmxt_tmia): Likewise.
4987         (iwmmxt_tmiaph): Likewise.
4988         (iwmmxt_tmiabb): Likewise.
4989         (iwmmxt_tmiatb): Likewise.
4990         (iwmmxt_tmiabt): Likewise.
4991         (iwmmxt_tmiatt): Likewise.
4992         (iwmmxt_tmovmskb): Likewise.
4993         (iwmmxt_tmovmskh): Likewise.
4994         (iwmmxt_tmovmskw): Likewise.
4995         (iwmmxt_waccb): Likewise.
4996         (iwmmxt_wacch): Likewise.
4997         (iwmmxt_waccw): Likewise.
4998         (iwmmxt_waligni): Likewise.
4999         (iwmmxt_walignr): Likewise.
5000         (iwmmxt_walignr0): Likewise.
5001         (iwmmxt_walignr1): Likewise.
5002         (iwmmxt_walignr2): Likewise.
5003         (iwmmxt_walignr3): Likewise.
5004         (iwmmxt_wsadb): Likewise.
5005         (iwmmxt_wsadh): Likewise.
5006         (iwmmxt_wsadbz): Likewise.
5007         (iwmmxt_wsadhz): Likewise.
5008         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5009         (iwmmxt_wabsdiffb): Likewise.
5010         (iwmmxt_wabsdiffh): Likewise.
5011         (iwmmxt_wabsdiffw): Likewise.
5012         (iwmmxt_waddsubhx): Likewise
5013         (iwmmxt_wsubaddhx): Likewise.
5014         (addc<mode>3): Likewise.
5015         (iwmmxt_avg4): Likewise.
5016         (iwmmxt_avg4r): Likewise.
5017         (iwmmxt_wmaddsx): Likewise.
5018         (iwmmxt_wmaddux): Likewise.
5019         (iwmmxt_wmaddsn): Likewise.
5020         (iwmmxt_wmaddun): Likewise.
5021         (iwmmxt_wmulwsm): Likewise.
5022         (iwmmxt_wmulwum): Likewise.
5023         (iwmmxt_wmulsmr): Likewise.
5024         (iwmmxt_wmulumr): Likewise.
5025         (iwmmxt_wmulwsmr): Likewise.
5026         (iwmmxt_wmulwumr): Likewise.
5027         (iwmmxt_wmulwl): Likewise.
5028         (iwmmxt_wqmulm): Likewise.
5029         (iwmmxt_wqmulwm): Likewise.
5030         (iwmmxt_wqmulmr): Likewise.
5031         (iwmmxt_wqmulwmr): Likewise.
5032         (iwmmxt_waddbhusm): Likewise.
5033         (iwmmxt_waddbhusl): Likewise.
5034         (iwmmxt_wqmiabb): Likewise.
5035         (iwmmxt_wqmiabt): Likewise.
5036         (iwmmxt_wqmiatb): Likewise.
5037         (iwmmxt_wqmiatt): Likewise.
5038         (iwmmxt_wqmiabbn): Likewise.
5039         (iwmmxt_wqmiabtn): Likewise.
5040         (iwmmxt_wqmiatbn): Likewise.
5041         (iwmmxt_wqmiattn): Likewise.
5042         (iwmmxt_wmiabb): Likewise.
5043         (iwmmxt_wmiabt): Likewise.
5044         (iwmmxt_wmiatb): Likewise.
5045         (iwmmxt_wmiatt): Likewise.
5046         (iwmmxt_wmiabbn): Likewise.
5047         (iwmmxt_wmiabtn): Likewise.
5048         (iwmmxt_wmiatbn): Likewise.
5049         (iwmmxt_wmiattn): Likewise.
5050         (iwmmxt_wmiawbb): Likewise.
5051         (iwmmxt_wmiawbt): Likewise.
5052         (iwmmxt_wmiawtb): Likewise.
5053         (iwmmxt_wmiawtt): Likewise.
5054         (iwmmxt_wmiawbbn): Likewise.
5055         (iwmmxt_wmiawbtn): Likewise.
5056         (iwmmxt_wmiawtbn): Likewise.
5057         (iwmmxt_wmiawttn): Likewise.
5058         (iwmmxt_wmerge): Likewise.
5059         (iwmmxt_tandc<mode>3): Likewise.
5060         (iwmmxt_torc<mode>3): Likewise.
5061         (iwmmxt_torvsc<mode>3): Likewise.
5062         (iwmmxt_textrc<mode>3): Likewise.
5063         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5064         (wmmxt_pack): Likewise.
5065         (wmmxt_mult_c1): Likewise.
5066         (wmmxt_mult_c2): Likewise.
5067         (wmmxt_alu_c1): Likewise.
5068         (wmmxt_alu_c2): Likewise.
5069         (wmmxt_alu_c3): Likewise.
5070         (wmmxt_transfer_c1): Likewise.
5071         (wmmxt_transfer_c2): Likewise.
5072         (wmmxt_transfer_c3): Likewise.
5073         (marvell_f_iwmmxt_wstr): Likewise.
5074         (marvell_f_iwmmxt_wldr): Likewise.
5076 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5078         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5080 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5082         Revert:
5083         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5084         * lra-constraints.c (need_for_split_p): Check call used hard regs
5085         living through calls.
5087         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5088         call used regs for call insn.
5090 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5092         PR target/57736
5093         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5094         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5096 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5098         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5100 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5102         * lra-constraints.c (need_for_split_p): Check call used hard regs
5103         living through calls.
5105 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
5107         PR target/57744
5108         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5109         to tie with any other modes.  Eliminate Altivec vector mode tests,
5110         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
5111         code, to return 0 if testing MODE2 for a condition, if we've
5112         already tested MODE1 for the same condition.
5114 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5116         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5117         layout.
5119 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5121         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5122         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5124 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5126         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5127         Define.
5128         (aarch64_symbolic_constant_p): Remove.
5129         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5130         static.  Fix line length and white space.
5131         (aarch64_symbolic_constant_p): Remove.
5132         * config/aarch64/predicates.md (aarch64_valid_symref):
5133         Use aarch64_classify_symbol_expression.
5135 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5137         * config/arm/constraints.md (Ts): New constraint.
5138         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5139         16-bit encodings.
5140         (compare_scc): Use "Ts" constraint for operand 0.
5141         (ior_scc_scc): Likewise.
5142         (and_scc_scc): Likewise.
5143         (and_scc_scc_nodom): Likewise.
5144         (ior_scc_scc_cmp): Likewise for operand 7.
5145         (and_scc_scc_cmp): Likewise.
5146         * config/arm/thumb2.md (thumb2_movsi_insn):
5147         Add alternatives for 16-bit encodings.
5148         (thumb2_movhi_insn): Likewise.
5149         (thumb2_movsicc_insn): Likewise.
5150         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
5151         (thumb2_negscc): Use "Ts" constraint.
5152         Move mvn instruction outside cond_exec block.
5153         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5154         for 16-bit encodings.
5156 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5158         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5159         encoding.
5160         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5161         (mulsi3subsi): Likewise.
5162         (mulsidi3adddi): Likewise.
5163         (mulsidi3_v6): Likewise.
5164         (umulsidi3_v6): Likewise.
5165         (umulsidi3adddi_v6): Likewise.
5166         (smulsi3_highpart_v6): Likewise.
5167         (umulsi3_highpart_v6): Likewise.
5168         (mulhisi3tb): Likewise.
5169         (mulhisi3bt): Likewise.
5170         (mulhisi3tt): Likewise.
5171         (maddhisi4): Likewise.
5172         (maddhisi4tb): Likewise.
5173         (maddhisi4tt): Likewise.
5174         (maddhidi4): Likewise.
5175         (maddhidi4tb): Likewise.
5176         (maddhidi4tt): Likewise.
5177         (zeroextractsi_compare0_scratch): Likewise.
5178         (insv_zero): Likewise.
5179         (insv_t2): Likewise.
5180         (anddi_notzesidi_di): Likewise.
5181         (anddi_notsesidi_di): Likewise.
5182         (andsi_notsi_si): Likewise.
5183         (iordi_zesidi_di): Likewise.
5184         (xordi_zesidi_di): Likewise.
5185         (andsi_iorsi3_notsi): Likewise.
5186         (smax_0): Likewise.
5187         (smax_m1): Likewise.
5188         (smin_0): Likewise.
5189         (not_shiftsi): Likewise.
5190         (unaligned_loadsi): Likewise.
5191         (unaligned_loadhis): Likewise.
5192         (unaligned_loadhiu): Likewise.
5193         (unaligned_storesi): Likewise.
5194         (unaligned_storehi): Likewise.
5195         (extv_reg): Likewise.
5196         (extzv_t2): Likewise.
5197         (divsi3): Likewise.
5198         (udivsi3): Likewise.
5199         (arm_zero_extendhisi2addsi): Likewise.
5200         (arm_zero_extendqisi2addsi): Likewise.
5201         (compareqi_eq0): Likewise.
5202         (arm_extendhisi2_v6): Likewise.
5203         (arm_extendqisi2addsi): Likewise.
5204         (arm_movt): Likewise.
5205         (thumb2_ldrd): Likewise.
5206         (thumb2_ldrd_base): Likewise.
5207         (thumb2_ldrd_base_neg): Likewise.
5208         (thumb2_strd): Likewise.
5209         (thumb2_strd_base): Likewise.
5210         (thumb2_strd_base_neg): Likewise.
5211         (arm_negsi2): Add alternative for 16-bit encoding.
5212         (arm_one_cmplsi2): Likewise.
5214 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5216         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5217         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5218         (movdfcc): Likewise.
5219         * config/arm/vfp.md (*thumb2_movsf_vfp):
5220         Disable predication for arm_restrict_it.
5221         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5222         (*thumb2_movdfcc_vfp): Likewise.
5223         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5224         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5225         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5226         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5227         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5228         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5229         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5230         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5231         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5232         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5233         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5234         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5235         Disable predication for arm_restrict_it.
5237 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
5239         * config/i386/bmiintrin.h (_bextr_u32): New.
5240         (_bextr_u64): Ditto.
5242 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
5244         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5245         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5246         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5247         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5248         * config/mips/n32-elf.h: ...this new file.
5250 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
5252         PR target/57224
5253         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5254         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5256 2013-06-27  Catherine Moore <clm@codesourcery.com>
5258         * config/mips/mips-tables.opt: Regenerate.
5259         * config/mips/mips-cpus.def: Add m14ke and m14kec.
5260         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5261         * doc/invoke.texi: Add -m14kc.
5263 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5265         PR target/57623
5266         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5267         constraints of operand 1 and 2.
5269         PR target/57623
5270         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5271         to match RTL canonicalization.  Swap predicates and
5272         constraints of operand 1 and 2.
5274 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
5276         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5277         Process OP_INOUT regs for splitting too.
5279 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5281         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5282         decl before the loop, initialize to NULL.
5283         (vectorizable_load): Initialize ptr_incr to NULL.
5285 2013-06-27  Martin Jambor  <mjambor@suse.cz>
5287         PR lto/57208
5288         * ipa-ref.h (ipa_maybe_record_reference): Declare.
5289         * ipa-ref.c (ipa_maybe_record_reference): New function.
5290         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5291         * ipa-cp.c (create_specialized_node): Record potential references from
5292         aggvals.
5293         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5295 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5297         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5298         parameter 'mode' of type 'enum machine_mode mode'; change to pass
5299         'mode' to force_reg.
5300         (aarch64_add_offset): Update calls to aarch64_force_temporary.
5301         (aarch64_expand_mov_immediate): Likewise.
5303 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5305         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
5306         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
5308 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5310         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
5311         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5312         (struct machine_function): Add tbegin_p.
5313         (s390_canonicalize_comparison): Fold CC mode compares to
5314         conditional jump if possible.
5315         (s390_emit_jump): Return the emitted jump.
5316         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
5317         Handle CCRAWmode compares.
5318         (s390_option_override): Default to -mhtm if available.
5319         (s390_reg_clobbered_rtx): Handle floating point regs as well.
5320         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
5321         FPRs instead of df_regs_ever_live_p.
5322         (s390_optimize_nonescaping_tx): New function.
5323         (s390_init_frame_layout): Extend clobbered_regs array to cover
5324         FPRs as well.
5325         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
5326         (s390_expand_tbegin): New function.
5327         (enum s390_builtin): New enum definition.
5328         (code_for_builtin): New array definition.
5329         (s390_init_builtins): New function.
5330         (s390_expand_builtin): New function.
5331         (TARGET_INIT_BUILTINS): Define.
5332         (TARGET_EXPAND_BUILTIN): Define.
5333         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
5334         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
5335         (s390_alc_comparison): Likewise.
5336         * config/s390/s390-modes.def: Add CCRAWmode.
5337         * config/s390/s390.h (processor_flags): Add PF_TX.
5338         (TARGET_CPU_HTM): Define macro.
5339         (TARGET_HTM): Define macro.
5340         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
5341         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
5342         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5343         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
5344         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
5345         (TBEGIN_MASK, TBEGINC_MASK): New constants.
5346         ("*cc_to_int"): Move up.
5347         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
5348         constants other than 0.
5349         ("*ccraw_to_int"): New insn and splitter definition.
5350         ("tbegin", "tbegin_nofloat", "tbegin_retry")
5351         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
5352         ("tx_assist"): New expander.
5353         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
5354         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
5355         * config/s390/s390.opt: Add -mhtm option.
5356         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
5357         * config/s390/htmxlintrin.h: New file.
5358         * config/s390/htmintrin.h: New file.
5359         * config/s390/s390intrin.h: New file.
5360         * doc/extend.texi: Document htm builtins.
5361         * config.gcc: Add the new header files to extra_headers.
5363 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
5365         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
5366         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
5368 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
5369             Pat Haugen <pthaugen@us.ibm.com>
5370             Peter Bergner <bergner@vnet.ibm.com>
5372         * config/rs6000/power8.md: New.
5373         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
5374         setting for power8 entry.
5375         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
5376         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
5377         test for Power4/Power5 only.
5378         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
5379         support.
5380         (force_new_group): Adjust comment.
5381         * config/rs6000/rs6000.md: Include power8.md.
5383 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
5385         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
5386         * config/arm/arm-protos.h (arm_max_conditional_execute): New
5387         declaration.
5388         (tune_params): Update comment.
5389         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
5390         (arm_max_conditional_execute): New function.
5391         (thumb2_final_prescan_insn): Use max_insn_skipped and
5392         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
5394 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
5396         PR tree-optimization/57705
5397         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
5398         SSA_NAME step, provided that it is not defined inside the loop.
5399         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
5400         (get_initial_def_for_induction): Handle SSA_NAME IV step.
5402 2013-06-25  Martin Jambor  <mjambor@suse.cz>
5404         PR middle-end/57670
5405         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
5406         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
5407         calls in the dump.
5408         (ipa_note_param_call): Initialize member_ptr flag.
5409         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
5410         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
5411         (ipa_write_indirect_edge_info): Stream member_ptr flag.
5412         (ipa_read_indirect_edge_info): Likewise.
5414 2013-06-25  Richard Biener  <rguenther@suse.de>
5416         PR middle-end/56977
5417         * passes.c (init_optimization_passes): Move pass_fold_builtins
5418         and pass_dce earlier with -Og.
5420 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
5422         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
5423         <BIT_FIELD_REF>: Remove trailing TAB.
5424         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
5425         remove blank line.
5427 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5429         PR tree-optimization/57358
5430         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
5431         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
5432         (ipa_analyze_params_uses): Generate pessimistic info when true.
5434 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5436         PR tree-optimization/57539
5437         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
5438         global.inlined_to of the new node to it.  All callers changed.
5439         * ipa-inline-transform.c (clone_inlined_nodes): New variable
5440         inlining_into, pass it to cgraph_clone_node.
5441         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
5442         ipa_free_edge_args_substructures.
5443         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
5444         rdesc linked list.  Do not assert rdesc edges have inlined caller.
5445         Assert we have found an rdesc in the rdesc list.
5447 2013-06-24  Richard Biener  <rguenther@suse.de>
5449         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
5450         (pointer_set_lookup): Declare.
5451         (class pointer_map): New template class implementing a
5452         generic pointer to T map.
5453         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
5454         pointer_map<T>::contains, pointer_map<T>::insert,
5455         pointer_map<T>::traverse): New functions.
5456         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
5457         (pointer_set_lookup): New function.
5458         (pointer_set_contains): Use pointer_set_lookup.
5459         (pointer_set_insert): Likewise.
5460         (insert_aux): Remove.
5461         (struct pointer_map_t): Embed a pointer_set_t.
5462         (pointer_map_create): Adjust.
5463         (pointer_map_destroy): Likewise.
5464         (pointer_map_contains): Likewise.
5465         (pointer_map_insert): Likewise.
5466         (pointer_map_traverse): Likewise.
5467         * tree-streamer.h (struct streamer_tree_cache_d): Use a
5468         pointer_map<unsigned> instead of a pointer_map_t.
5469         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
5470         (streamer_tree_cache_lookup): Likewise.
5471         (streamer_tree_cache_create): Likewise.
5472         (streamer_tree_cache_delete): Likewise.
5473         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
5474         pointer_map<unsigned> instead of a pointer_map_t.
5475         (lto_init_tree_ref_encoder): Adjust.
5476         (lto_destroy_tree_ref_encoder): Likewise.
5477         * lto-section-out.c (lto_output_decl_index): Likewise.
5478         (lto_record_function_out_decl_state): Likewise.
5479         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
5481 2013-06-24  Richard Biener  <rguenther@suse.de>
5483         PR tree-optimization/57488
5484         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
5486 2013-06-24  Alan Modra  <amodra@gmail.com>
5488         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
5489         (gen_easy_altivec_constant): Likewise.
5490         * config/rs6000/predicates.md (easy_vector_constant_add_self,
5491         easy_vector_constant_msb): Likewise.
5493 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
5495         PR target/57688
5496         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
5497         add missing return true.
5499 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
5501         PR target/52483
5502         * config/sh/predicates.md (general_extend_operand): Invoke
5503         general_movsrc_operand for memory operands.
5504         (general_movsrc_operand): Allow reg+reg addressing, do not use
5505         general_operand for memory operands.
5507 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
5509         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
5510         when current target options does not apply.
5511         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
5512         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
5513         * config/i386/bmiintrin.h: Pass appropriate target
5514         attributes to header.
5515         * config/i386/mmintrin.h: Ditto.
5516         * config/i386/nmmintrin.h: Ditto.
5517         * config/i386/avx2intrin.h: Ditto.
5518         * config/i386/fxsrintrin.h: Ditto.
5519         * config/i386/tbmintrin.h: Ditto.
5520         * config/i386/xsaveintrin.h: Ditto.
5521         * config/i386/f16cintrin.h: Ditto.
5522         * config/i386/xtestintrin.h: Ditto.
5523         * config/i386/xsaveoptintrin.h: Ditto.
5524         * config/i386/bmi2intrin.h: Ditto.
5525         * config/i386/lzcntintrin.h: Ditto.
5526         * config/i386/smmintrin.h: Ditto.
5527         * config/i386/wmmintrin.h: Ditto.
5528         * config/i386/x86intrin.h: Remove all header include guards.
5529         * config/i386/prfchwintrin.h: Ditto.
5530         * config/i386/pmmintrin.h: Ditto.
5531         * config/i386/tmmintrin.h: Ditto.
5532         * config/i386/xmmintrin.h: Ditto.
5533         * config/i386/popcntintrin.h: Ditto.
5534         * config/i386/rdseedintrin.h: Ditto.
5535         * config/i386/ammintrin.h: Ditto.
5536         * config/i386/emmintrin.h: Ditto.
5537         * config/i386/immintrin.h: Remove all header include guards.
5538         * config/i386/fma4intrin.h: Ditto.
5539         * config/i386/lwpintrin.h: Ditto.
5540         * config/i386/xopintrin.h: Ditto.
5541         * config/i386/ia32intrin.h: Ditto.
5542         * config/i386/avxintrin.h: Ditto.
5543         * config/i386/rtmintrin.h: Ditto.
5544         * config/i386/fmaintrin.h: Ditto.
5545         * config/i386/mm3dnow.h: Ditto.
5547 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
5549         * common/config/i386/i386-common.c: Handle LZCNT.
5551 2013-06-22  Andi Kleen  <ak@linux.intel.com>
5553         * doc/extend.texi: Use __atomic_store_n instead of
5554         __atomic_store in HLE example.
5556 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
5558         * config/sh/sh.c: Remove <cstdlib> workaround.
5560 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5562         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
5564 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5566         * doc/extend.texi: Document that __atomic_clear and
5567           __atomic_test_and_set should only be used with bool.
5569 2013-06-20  Jan Hubicka  <jh@suse.cz>
5571         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5572         types_same_for_odr.
5573         * tree.c (decls_same_for_odr): New function.
5574         (same_for_edr): New function.
5575         (types_same_for_odr): New function.
5576         (get_binfo_at_offset): Use it.
5577         * tree.h (types_same_for_odr): Declare.
5579 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
5580             Jason Merrill  <jason@redhat.com>
5582         * system.h: Include <cstdlib> as well as <stdlib.h>.
5584 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
5586         PR target/57655
5587         * config/i386/i386.c (construct_container): Report error if
5588         long double is used with disabled x87 float returns.
5590 2013-06-20  Jan Hubicka  <jh@suse.cz>
5592         * lto-cgraph.c (input_symtab): Do not set cgraph state.
5594 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
5596         PR rtl-optimization/57425
5597         PR rtl-optimization/57569
5598         * alias.c (write_dependence_p): Remove parameters mem_mode and
5599         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
5600         Changed all callers.
5601         (canon_anti_dependence): Get comments and semantics in sync.
5602         Add parameter mem_canonicalized.  Changed all callers.
5603         * rtl.h (canon_anti_dependence): Update prototype.
5605 2013-06-20  Richard Biener  <rguenther@suse.de>
5607         * data-streamer-in.c (streamer_read_uhwi): Optimize single
5608         byte case, inline streamer_read_uchar and defer section
5609         overrun check.
5611 2013-06-20  Richard Biener  <rguenther@suse.de>
5613         PR tree-optimization/57584
5614         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5615         SSA names into the expanded expression that take part in
5616         abnormal coalescing.
5618 2013-06-19  Sharad Singhai  <singhai@google.com>
5620         * gcov.c (print_usage): Handle new option.
5621         (process_args): Ditto.
5622         (get_gcov_intermediate_filename): New function.
5623         (output_intermediate_file): New function.
5624         (output_gcov_file): New function
5625         (generate_results): Handle new option.
5626         (release_function): Relase demangled name.
5627         (read_graph_file): Handle demangled name.
5628         (output_lines): Ditto.
5629         * doc/gcov.texi: Document gcov intermediate format.
5631 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
5633         PR bootstrap/57604
5634         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5635         (lra_emit_add): Use the functions.  Add comment about Y as an
5636         address segment.
5638 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
5640         PR driver/57652
5641         * collect2.c (collect_atexit): New.
5642         (collect_exit): Delete.
5643         (main): Register collect_atexit with atexit.
5644         (collect_wait): Change collect_exit to exit.
5645         (do_wait): Same.
5646         * collect2.h (collect_exit): Delete.
5647         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5649 2013-06-19  Wei Mi  <wmi@google.com>
5651         PR rtl-optimization/57518
5652         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5653         if regno is used in paradoxical subreg.
5654         (update_equiv_regs): Check pdx_subregs[regno] before
5655         set a reg to be equivalent with a mem.
5657 2013-06-19  Matthias Klose  <doko@ubuntu.com>
5659         PR driver/57651
5660         * file-find.h (find_a_file): Add a mode parameter.
5661         * file-find.c (find_a_file): Likewise.
5662         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5663         with X_OK for the executables.
5664         * collect2.c (main): Call find_a_file with X_OK.
5666 2013-06-19  Steve Ellcey  <sellcey@mips.com>
5668         PR target/56942
5669         * config/mips/mips.md (casesi_internal_mips16_<mode>):
5670         Use NEXT_INSN instead of next_real_insn.
5672 2013-06-19  Jan Hubicka  <jh@suse.cz>
5674         * cgraph.h (const_value_known_p): Replace by ...
5675         (ctor_for_folding): .. this one.
5676         * cgraphunit.c (process_function_and_variable_attributes): Use it.
5677         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5678         * expr.c (expand_expr_real_1): Likewise.
5679         (string_constant): Likewise.
5680         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5681         * ipa.c (process_references): Likewise.
5682         (symtab_remove_unreachable_nodes): Likewise.
5683         * ipa-inline-analysis.c (param_change_prob): Likewise.
5684         * gimple-fold.c (canonicalize_constructor_val): Likewise.
5685         (get_base_constructor): Likwise.
5686         * varpool.c (varpool_remove_node): Likewise.
5687         (varpool_remove_initializer): LIkewise.
5688         (dump_varpool_node): LIkwise.
5689         (const_value_known_p): Rewrite to ...
5690         (ctor_for_folding): ... this one.
5692 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
5694         PR driver/57651
5695         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5696         PERSONALITY in $PATH derived prefixes.
5698 2013-06-19  Jeff Law  <law@redhat.com>
5700         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5701         in comment.
5703         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5704         (simplify_bitwise_binary): Use it to simpify certain binary ops on
5705         booleans.
5707 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
5709         * config/arm/vfp.md: Move VFP instruction classification documentation
5710         to ...
5711         * config/arm/arm.md: ... here.  Update instruction classification
5712         documentation.
5714 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
5716         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
5717         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
5718         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
5720 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
5722         PR target/57609
5723         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5724         with NEXT_INSN.  Use tablejump_p to check for jump table data
5725         insns.
5727 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
5729         PR c++/56544
5730         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
5731         that now in C++ the value is correct per the C++ standards.
5733 2013-06-19  Richard Biener  <rguenther@suse.de>
5735         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
5736         for global context.
5738 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5740         Revert:
5741         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5743         PR target/57609
5744         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5745         with next_active_insn.
5747 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
5749         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
5750         functions are inlined during failures to flag an error.
5751         * tree-inline.c (expand_call_inline): Allow the error to be flagged
5752         in early inline pass.
5754 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
5756         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
5757         in comments.
5759 2013-06-18  Julian Brown  <julian@codesourcery.com>
5761         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
5762         Permit virtual register pre-reload if !strict.
5763         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
5764         change.
5765         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
5766         prototype.
5767         * config/arm/neon.md (movmisalign<mode>): Use
5768         neon_perm_struct_or_reg_operand instead of
5769         neon_struct_or_register_operand.
5770         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
5771         neon_permissive_struct_operand instead of neon_struct_operand.
5772         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
5773         neon_vector_mem_operand.
5774         * config/arm/predicates.md (neon_struct_operand): Adjust call to
5775         neon_vector_mem_operand.
5776         (neon_permissive_struct_operand): New.
5777         (neon_struct_or_register_operand): Rename to...
5778         (neon_perm_struct_or_reg_operand): This. Adjust call to
5779         neon_vector_mem_operand.
5781 2013-06-18  Richard Biener  <rguenther@suse.de>
5783         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
5784         * lto-streamer.h: Include pointer-set.h.
5785         (struct lto_decl_slot): Remove.
5786         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
5787         Remove next_index entry.
5788         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5789         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5790         (lto_init_tree_ref_encoder): Adjust.
5791         (lto_destroy_tree_ref_encoder): Likewise.
5792         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5793         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5794         (lto_output_decl_index): Adjust.
5795         (lto_new_out_decl_state): Likewise.
5796         (lto_record_function_out_decl_state): Likewise.
5797         * lto-streamer-out.c (copy_function): Likewise.
5799 2013-06-18  Richard Biener  <rguenther@suse.de>
5801         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5802         * cgraphunit.c: Include cfgloop.h.
5803         (init_lowered_empty_function): Initialize the loop tree.
5804         (assemble_thunk): Insert new BBs into loops.
5806 2013-06-18  Richard Biener  <rguenther@suse.de>
5808         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5809         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5810         the map from cache entry to cache index optional.
5811         (streamer_tree_cache_replace_tree): Adjust accordingly.
5812         (streamer_tree_cache_append): Likewise.
5813         (streamer_tree_cache_delete): Likewise.
5814         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5815         streamer cache map from cache entry to cache index.
5816         * lto-streamer-out.c (create_output_block): Adjust.
5818 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
5820         * config/arm/arm.md (attribute "insn"): Move multiplication and
5821         division attributes to...
5822         (attribute "type"): ... here.  Remove mult.
5823         (attribute "mul32"): New attribute.
5824         (attribute "mul64"): Add umaal.
5825         (*arm_mulsi3): Update attributes.
5826         (*arm_mulsi3_v6): Likewise.
5827         (*thumb_mulsi3): Likewise.
5828         (*thumb_mulsi3_v6): Likewise.
5829         (*mulsi3_compare0): Likewise.
5830         (*mulsi3_compare0_v6): Likewise.
5831         (*mulsi_compare0_scratch): Likewise.
5832         (*mulsi_compare0_scratch_v6): Likewise.
5833         (*mulsi3addsi): Likewise.
5834         (*mulsi3addsi_v6): Likewise.
5835         (*mulsi3addsi_compare0): Likewise.
5836         (*mulsi3addsi_compare0_v6): Likewise.
5837         (*mulsi3addsi_compare0_scratch): Likewise.
5838         (*mulsi3addsi_compare0_scratch_v6): Likewise.
5839         (*mulsi3subsi): Likewise.
5840         (*mulsidi3adddi): Likewise.
5841         (*mulsi3addsi_v6): Likewise.
5842         (*mulsidi3adddi_v6): Likewise.
5843         (*mulsidi3_nov6): Likewise.
5844         (*mulsidi3_v6): Likewise.
5845         (*umulsidi3_nov6): Likewise.
5846         (*umulsidi3_v6): Likewise.
5847         (*umulsidi3adddi): Likewise.
5848         (*umulsidi3adddi_v6): Likewise.
5849         (*smulsi3_highpart_nov6): Likewise.
5850         (*smulsi3_highpart_v6): Likewise.
5851         (*umulsi3_highpart_nov6): Likewise.
5852         (*umulsi3_highpart_v6): Likewise.
5853         (mulhisi3): Likewise.
5854         (*mulhisi3tb): Likewise.
5855         (*mulhisi3bt): Likewise.
5856         (*mulhisi3tt): Likewise.
5857         (maddhisi4): Likewise.
5858         (*maddhisi4tb): Likewise.
5859         (*maddhisi4tt): Likewise.
5860         (maddhidi4): Likewise.
5861         (*maddhidi4tb): Likewise.
5862         (*maddhidi4tt): Likewise.
5863         (divsi3): Likewise.
5864         (udivsi3): Likewise.
5865         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5866         (thumb2_mulsi_short_compare0): Likewise.
5867         (thumb2_mulsi_short_compare0_scratch): Likewise.
5868         * config/arm/arm1020e.md (1020mult1): Update attribute change.
5869         (1020mult2): Likewise.
5870         (1020mult3): Likewise.
5871         (1020mult4): Likewise.
5872         (1020mult5): Likewise.
5873         (1020mult6): Likewise.
5874         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5875         change.
5876         (cortex_a15_mult64): Likewise.
5877         (cortex_a15_sdiv): Likewise.
5878         (cortex_a15_udiv): Likewise.
5879         * config/arm/arm1026ejs.md (mult1): Update attribute change.
5880         (mult2): Likewise.
5881         (mult3): Likewise.
5882         (mult4): Likewise.
5883         (mult5): Likewise.
5884         (mult6): Likewise.
5885         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5886         (pj4_ir_div): Likewise.
5887         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5888         (11_mult2): Likewise.
5889         (11_mult3): Likewise.
5890         (11_mult4): Likewise.
5891         (11_mult5): Likewise.
5892         (11_mult6): Likewise.
5893         (11_mult7): Likewise.
5894         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5895         (cortex_a8_mla): Likewise.
5896         (cortex_a8_mull): Likewise.
5897         (cortex_a8_smulwy): Likewise.
5898         (cortex_a8_smlald): Likewise.
5899         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5900         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5901         (cortex_r4_mul_3): Likewise.
5902         (cortex_r4_mla_4): Likewise.
5903         (cortex_r4_mla_3): Likewise.
5904         (cortex_r4_smlald): Likewise.
5905         (cortex_r4_mull): Likewise.
5906         (cortex_r4_sdiv): Likewise.
5907         (cortex_r4_udiv): Likewise.
5908         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5909         (cortex_a7_idiv): Likewise.
5910         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5911         (9_mult2): Likewise.
5912         (9_mult3): Likewise.
5913         (9_mult4): Likewise.
5914         (9_mult5): Likewise.
5915         (9_mult6): Likewise.
5916         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5917         (cortex_a53_sdiv): Likewise.
5918         (cortex_a53_udiv): Likewise.
5919         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5920         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5921         (mp626_mult2): Likewise.
5922         (mp626_mult3): Likewise.
5923         (mp626_mult4): Likewise.
5924         * config/arm/fa526.md (526_mult1): Update attribute change.
5925         (526_mult2): Likewise.
5926         * config/arm/arm-generic.md (mult): Update attribute change.
5927         (mult_ldsched_strongarm): Likewise.
5928         (mult_ldsched): Likewise.
5929         (multi_cycle): Likewise.
5930         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5931         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5932         (606te_mult2): Likewise.
5933         (606te_mult3): Likewise.
5934         (606te_mult4): Likewise.
5935         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5936         (cortex_a9_mac16): Likewise.
5937         (cortex_a9_multiply): Likewise.
5938         (cortex_a9_mac): Likewise.
5939         (cortex_a9_multiply_long): Likewise.
5940         * config/arm/fa626te.md (626te_mult1): Update attribute change.
5941         (626te_mult2): Likewise.
5942         (626te_mult3): Likewise.
5943         (626te_mult4): Likewise.
5945 2013-06-18  Richard Biener  <rguenther@suse.de>
5947         PR lto/57334
5948         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5950 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5952         PR target/57609
5953         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5954         with next_active_insn.
5956 2013-06-18  Alan Modra  <amodra@gmail.com>
5958         * config/rs6000/rs6000.h (enum data_align): New.
5959         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5960         (DATA_ABI_ALIGNMENT): Define.
5961         (CONSTANT_ALIGNMENT): Correct comment.
5962         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5963         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5965 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5967         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5968         ATTRIBUTE_UNUSED marking.
5970 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5972         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5973         alternative and update.
5974         (aarch64_dup_lanedi): Delete.
5975         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5976         * config/aarch64/aarch64-simd-builtins.def: Update.
5978 2013-06-17  Richard Biener  <rguenther@suse.de>
5980         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5981         (lto_input_scc): Declare.
5982         (lto_input_tree_1): Likewise.
5983         (struct lto_stats_d): Add num_tree_bodies_output and
5984         num_pickle_refs_output.
5985         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5986         (lto_read_tree_1): Split out from ...
5987         (lto_read_tree): ... this.
5988         (lto_input_scc): New function.
5989         (lto_input_tree_1): Split out from ...
5990         (lto_input_tree): ... this.  Handle LTO_tree_scc.
5991         (lto_data_in_create): Create the streamer cache without hashes.
5992         * lto-streamer-out.c (create_output_block): Create the streamer
5993         cache with hashes when not doing WPA.
5994         (lto_write_tree_1): Split out from ...
5995         (lto_write_tree): ... this.
5996         (get_symbol_initial_value): New function.
5997         (lto_output_tree_1): Split out from ...
5998         (lto_output_tree): ... this.  Write trees as series of SCCs
5999         using a DFS walk via DFS_write_tree.
6000         (struct sccs, struct scc_entry): New types.
6001         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6002         (DFS_write_tree_body): New function.
6003         (DFS_write_tree): Likewise.
6004         (hash_tree): Likewise.
6005         (scc_entry_compare): Likewise.
6006         (hash_scc): Likewise.
6007         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6008         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6009         TREE_CHAIN as regular reference.
6010         (streamer_read_integer_cst): Remove.
6011         (streamer_get_pickled_tree): Adjust.
6012         * tree-streamer-out.c (streamer_write_chain): Disable streaming
6013         of DECL_EXTERNALs in BLOCK_VARS for now.
6014         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6015         reference.
6016         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6017         Add hash value argument and record that if hashes are recorded
6018         in the cache.
6019         (streamer_tree_cache_insert_1): Adjust.
6020         (streamer_tree_cache_insert): Likewise.
6021         (streamer_tree_cache_insert_at): Rename to ...
6022         (streamer_tree_cache_replace_tree): ... this and adjust.
6023         (streamer_tree_cache_append): Adjust.
6024         (record_common_node): Likewise.
6025         (streamer_tree_cache_create): Add argument whether to
6026         record hash values together with trees.
6027         (streamer_tree_cache_delete): Adjust.
6028         * tree-streamer.h (struct streamer_tree_cache_d): Add
6029         vector of hashes.
6030         (streamer_read_integer_cst): Remove.
6031         (streamer_tree_cache_insert): Adjust.
6032         (streamer_tree_cache_append): Likewise.
6033         (streamer_tree_cache_insert_at): Rename to ...
6034         (streamer_tree_cache_replace_tree): ... this and adjust.
6035         (streamer_tree_cache_create): Add argument whether to record hashes.
6036         (streamer_tree_cache_get): Rename to ...
6037         (streamer_tree_cache_get_tree): ... this.
6038         (streamer_tree_cache_get_hash): New function.
6039         * tree.c (cache_integer_cst): New function.
6040         * tree.h (cache_integer_cst): Declare.
6041         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6042         * lto-symtab.c (lto_varpool_replace_node): Only release
6043         DECL_INITIAL of non-prevailing decls.
6044         * varpool.c (varpool_remove_initializer): Do not release
6045         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6047 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6049         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6050         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6051         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6052         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6053         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6054         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6055         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6056         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6057         instead of TARGET_64BIT.
6058         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6059         Require ISA_HAS_<D>DIV.
6061 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6063         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6064         (mips*-*-linux*): Move default with_llsc setting to where other
6065         defaults are set.
6066         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6067         with_arch block.
6068         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6069         Likewise.  Remove default with_tune setting.  Move default float
6070         setting to its own block.  Handle with_llsc in the same block as above.
6072 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
6074         PR rtl-optimization/57425
6075         PR rtl-optimization/57569
6076         * alias.c (write_dependence_p): Add new parameters mem_mode,
6077         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6078         Changed all callers.
6079         (canon_anti_dependence): New function.
6080         * cse.c (check_dependence): Use canon_anti_dependence.
6081         * cselib.c (cselib_invalidate_mem): Likewise.
6082         * rtl.h (canon_anti_dependence): Declare.
6084 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6086         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6087         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6088         ".set mips3" for 64-bit targets.
6090 2013-06-15  Dehao Chen  <dehao@google.com>
6092         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6093         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6094         (gimple_check_call_args): Likewise.
6095         * value-prof.c (check_ic_target): Likewise.
6096         * ipa-inline.c (early_inliner): Likewise.
6097         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6098         * cgraph.c (cgraph_create_edge_1): Likewise.
6099         (cgraph_make_edge_direct): Likewise.
6101 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6103         PR target/57615
6104         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6105         rs6000_output_move_128bit to handle emitting quad memory
6106         operations.  Set attribute length to 8 bytes.
6108 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
6110         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6111         New pattern.
6112         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6113         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6114         (aarch64_<su>mlsl<mode>): Likewise.
6116 2013-06-14  Mike Stump  <mikestump@comcast.net>
6118         * Makefile.in (TARGET_H): Add insn-codes.h.
6120 2013-06-14  Alan Modra  <amodra@gmail.com>
6122         PR middle-end/57134
6123         PR middle-end/57586
6124         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6125         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
6126         bitfield expansion when EXPAND_MEMORY.
6127         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6129 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
6131         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6132         test for clearing quad memory on 32-bit later.
6134 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6136         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6137         (fold_negate_expr): Likewise.
6138         (fold_real_zero_addition_p): Handle vectors.
6139         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6141 2013-06-14  Alan Modra  <amodra@gmail.com>
6143         * varasm.c (force_const_mem): Revert 2013-06-07 change.
6145 2013-06-13  Jan Hubicka  <jh@suse.cz>
6147         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6148         Local comdats are not externally visible.
6149         * symtab.c (dump_symtab_base): Dump externally visible.
6150         (verify_symtab_base): Verify back links in the symtab hash.
6152 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6154         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6155         CONVERT_EXPR as equal nodes.
6157 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6159         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6161 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6163         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6164         Generalize to complex and vector.
6165         * tree.c (build_all_ones_cst): New function.
6166         * tree.h (build_all_ones_cst): Declare it.
6168 2013-06-13  Alan Modra  <amodra@gmail.com>
6170         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6171         * config/rs6000/rs6000.md (signbittf2): New insn.
6172         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6173         (abstf2_internal, cmptf_internal2): Likewise.
6174         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6176 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6177             Pat Haugen <pthaugen@us.ibm.com>
6178             Peter Bergner <bergner@vnet.ibm.com>
6180         * config/rs6000/rs6000.c (emit_load_locked): Add support for
6181         power8 byte, half-word, and quad-word atomic instructions.
6182         (emit_store_conditional): Likewise.
6183         (rs6000_expand_atomic_compare_and_swap): Likewise.
6184         (rs6000_expand_atomic_op): Likewise.
6186         * config/rs6000/sync.md (larx): Add new modes for power8.
6187         (stcx): Likewise.
6188         (AINT): New mode iterator to include TImode as well as normal
6189         integer modes on power8.
6190         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6191         that VSX registers are not considered.  Use AINT mode iterator
6192         instead of INT1 to allow inclusion of quad word atomic operations
6193         on power8.
6194         (load_locked<mode>): Likewise.
6195         (store_conditional<mode>): Likewise.
6196         (atomic_compare_and_swap<mode>): Likewise.
6197         (atomic_exchange<mode>): Likewise.
6198         (atomic_nand<mode>): Likewise.
6199         (atomic_fetch_<fetchop_name><mode>): Likewise.
6200         (atomic_nand_fetch<mode>): Likewise.
6201         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6202         each type.
6203         (ATOMIC): On power8, add QImode, HImode modes.
6204         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6205         modes that promote to SImode.
6206         (load_lockedti): Convert TImode arguments to PTImode, so that we
6207         get a guaranteed even/odd register pair.
6208         (load_lockedpti): Likewise.
6209         (store_conditionalti): Likewise.
6210         (store_conditionalpti): Likewise.
6212         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6213         atomic load/store instructions.
6214         (HSI): Likewise.
6216 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
6218         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6219         loads.
6220         (insn_count): New attribute, with most cases extracted from...
6221         (length): ...here.  Redefine most cases in terms of insn_count.
6222         (single_insn): Delete.
6223         (can_delay): Use insn_count to check for single instructions.
6224         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6225         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6226         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6227         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6228         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6229         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6230         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6231         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6232         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6233         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6234         rather than "length".
6235         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
6236         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6237         Use "insn_count" rather than "length".
6238         * config/mips/mips-dsp.md
6239         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6240         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6241         length attributes.
6243 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
6245         PR tree-optimization/57361
6246         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6248 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
6250         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6251         to split.
6252         (aarch64_simd_combine<mode>): New instruction expansion.
6253         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6254         function prototype.
6255         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6256         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6258 2013-06-12  Jan Hubicka  <jh@suse.cz>
6260         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6261         decl has when in streaming stage.
6262         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6263         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6265 2013-06-12  Roland Stigge <stigge@antcom.de>
6267         PR target/57578
6268         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6270 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6272         PR tree-optimization/57537
6273         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6274         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6276 2013-06-12  Richard Biener  <rguenther@suse.de>
6278         * data-streamer.h (streamer_write_char_stream): CSE
6279         obs->current_pointer.
6280         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6281         streamer_write_char_stream manually and optimize the resulting loop.
6282         (streamer_write_hwi_stream): Likewise.
6284 2013-06-12  Jan Hubicka  <jh@suse.cz>
6286         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6287         * cgraph.h (varpool_create_empty_node): Declare.
6288         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6289         duplicated nodes.
6290         * symtab.c (symtab_unregister_node): Be lax about missin entries
6291         in node hash.
6292         (symtab_get_node): Update comment.
6293         * varpool.c (varpool_create_empty_node): Break out from ...
6294         (varpool_node_for_decl): ... here.
6295         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6297 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
6299         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6300         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6301         part.  Use straight-line flow at the end.
6302         <COMPONENT_REF>: Remove superfluous else.
6303         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
6305 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6307         PR target/56564
6308         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
6309         target hook even for !TREE_PUBLIC decls.  If no resolution info
6310         is available, return false for common and external decls.
6312 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6314         * config/rl78/constraints.md (U): New constraint.
6315         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
6316         valloc attribute.
6318 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
6320         PR target/57589
6321         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
6322         to allow returning address to AT_PLATFORM name.
6324 2013-06-11  Jan Hubicka  <jh@suse.cz>
6326         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
6327         * cgraph.h (symtab_node_base): Add weakref flag.
6328         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
6329         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
6330         (output_weakrefs): Use weakref flag.
6331         * fold-const.c (simple_operand_p): Handle WEAK.
6332         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
6333         * ipa.c (varpool_externally_visible_p): Drop weakref.
6334         (function_and_variable_visibility): Update comment; fix weakref
6335         sanity checks; do not clear DECL_WEAK on them.
6336         * lto-cgraph.c (lto_output_node): update.
6337         (lto_output_varpool_node): Update.
6338         (input_overwrite_node): Update.
6339         (input_node): Update.
6340         (input_varpool_node): Update.
6341         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
6342         (lto_symtab_merge_symbols): Add sanity check.
6343         (lto_symtab_prevailing_decl): Do not special case weakrefs.
6344         * passes.c (rest_of_decl_compilation): Set static flag, too.
6345         * symtab.c (dump_symtab_base): Dump weakref.
6346         (verify_symtab_base): Sanity check weakrefs.
6347         (symtab_make_decl_local): Remove duplicated code.
6348         (symtab_alias_ultimate_target): Simplify.
6349         * varpool.c (varpool_create_variable_alias): Set weakref flag.
6351 2013-06-11  DJ Delorie  <dj@redhat.com>
6353         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
6354         (rl78_unwind_word_mode): New.
6356 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6358         * final.c (debug_prefix_maps): Make static.
6360 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6362         * function.c (initial_trampoline): Remove stray copy.
6364 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
6366         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
6368 2013-06-11  Martin Jambor  <mjambor@suse.cz>
6370         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
6371         within bounds at the beginning of the function.
6373 2013-06-11  Alan Modra  <amodra@gmail.com>
6375         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
6376         reporting.
6377         (get_named_section): Don't NULL !DECL_P decl.
6379 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
6381         * doc/invoke.texi (core-avx2): Document.
6382         (slm): Likewise.
6383         (atom): Updated with MOVBE.
6385 2013-06-11  Richard Biener  <rguenther@suse.de>
6387         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
6389 2013-06-11  Anton Blanchard  <anton@samba.org>
6391         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
6392         correct shift value in little-endian mode.
6394 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
6396         PR target/56564
6397         * varasm.c (get_variable_align): Move #endif to the right place.
6399 2013-06-10  Cary Coutant  <ccoutant@google.com>
6401         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
6402         for hash so that hash table traversal order is deterministic.
6404 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
6405             Pat Haugen <pthaugen@us.ibm.com>
6406             Peter Bergner <bergner@vnet.ibm.com>
6408         * config/rs6000/vector.md (GPR move splitter): Do not split moves
6409         of vectors in GPRS if they are direct moves or quad word load or
6410         store moves.
6412         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
6413         declaration.
6414         (direct_move_p): Likewise.
6415         (quad_load_store_p): Likewise.
6417         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
6418         classes into bins based on the physical register type.
6419         (reg_class_to_reg_type): Likewise.
6420         (IS_STD_REG_TYPE): Likewise.
6421         (IS_FP_VECT_REG_TYPE): Likewise.
6422         (reload_fpr_gpr): Arrays to determine what insn to use if we can
6423         use direct move instructions.
6424         (reload_gpr_vsx): Likewise.
6425         (reload_vsx_gpr): Likewise.
6426         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
6427         information that is a simplification of register classes.  Also
6428         precalculate direct move reload helpers.
6429         (direct_move_p): New function to return true if the operation can
6430         be done as a direct move instruciton.
6431         (quad_load_store_p): New function to return true if the operation
6432         is a quad memory operation.
6433         (rs6000_legitimize_address): If quad memory, only allow register
6434         indirect for TImode addresses.
6435         (rs6000_legitimate_address_p): Likewise.
6436         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
6437         (rs6000_reload_register_type): Likewise.
6438         (register_to_reg_type): Return register type.
6439         (rs6000_secondary_reload_simple_move): New helper function for
6440         secondary reload and secondary memory needed to identify anything
6441         that is a simple move, and does not need reloading.
6442         (rs6000_secondary_reload_direct_move): New helper function for
6443         secondary reload to identify cases that can be done with several
6444         instructions via the direct move instructions.
6445         (rs6000_secondary_reload_move): New helper function for secondary
6446         reload to identify moves between register types that can be done.
6447         (rs6000_secondary_reload): Add support for quad memory operations
6448         and for direct move.
6449         (rs6000_secondary_memory_needed): Likewise.
6450         (rs6000_debug_secondary_memory_needed): Change argument names.
6451         (rs6000_output_move_128bit): New function to return the move to
6452         use for 128-bit moves, including knowing about the various
6453         limitations of quad memory operations.
6455         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
6456         memory operations.  call rs6000_output_move_128bit for the actual
6457         instruciton(s) to generate.
6458         (vsx_movti_64bit): Likewise.
6460         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
6461         (UNSPEC_P8V_MTVSRWZ): Likewise.
6462         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
6463         (UNSPEC_P8V_MTVSRD): Likewise.
6464         (UNSPEC_P8V_XXPERMDI): Likewise.
6465         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
6466         (UNSPEC_FUSION_GPR): Likewise.
6467         (FMOVE128_GPR): New iterator for direct move.
6468         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
6469         (f32_sv): Likewise.
6470         (f32_dm): Likewise.
6471         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
6472         loads and direct move instructions.
6473         (zero_extendsidi2_lfiwzx): Likewise.
6474         (extendsidi2_lfiwax): Likewise.
6475         (extendsidi2_nocell): Likewise.
6476         (floatsi<mode>2_lfiwax): Likewise.
6477         (lfiwax): Likewise.
6478         (floatunssi<mode>2_lfiwzx): Likewise.
6479         (lfiwzx): Likewise.
6480         (fix_trunc<mode>_stfiwx): Likewise.
6481         (fixuns_trunc<mode>_stfiwx): Likewise.
6482         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
6483         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
6484         (parity<mode>2_cmpb): Set length/type attr.
6485         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
6486         for 'mr.' to fast_compare.
6487         (bpermd_<mode>): Change type attr to popcnt.
6488         (p8_fmrgow_<mode>): New insns for power8 direct move support.
6489         (p8_mtvsrwz_1): Likewise.
6490         (p8_mtvsrwz_2): Likewise.
6491         (reload_fpr_from_gpr<mode>): Likewise.
6492         (p8_mtvsrd_1): Likewise.
6493         (p8_mtvsrd_2): Likewise.
6494         (p8_xxpermdi_<mode>): Likewise.
6495         (reload_vsx_from_gpr<mode>): Likewise.
6496         (reload_vsx_from_gprsf): Likewise.
6497         (p8_mfvsrd_3_<mode>): LIkewise.
6498         (reload_gpr_from_vsx<mode>): Likewise.
6499         (reload_gpr_from_vsxsf): Likewise.
6500         (p8_mfvsrd_4_disf): Likewise.
6501         (multi-word GPR splits): Do not split direct moves or quad memory
6502         operations.
6504 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
6506         * tree-into-ssa.c (interesting_blocks): Make static.
6508 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
6510         PR target/56564
6511         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
6512         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
6513         Use DATA_ABI_ALIGNMENT for that case instead if defined.
6514         (get_variable_align): New function.
6515         (get_variable_section, emit_bss, emit_common,
6516         assemble_variable_contents, place_block_symbol): Use
6517         get_variable_align instead of DECL_ALIGN.
6518         (assemble_noswitch_variable): Add align argument, use it
6519         instead of DECL_ALIGN.
6520         (assemble_variable): Adjust caller.  Use get_variable_align
6521         instead of DECL_ALIGN.
6522         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
6523         caller.
6524         (DATA_ABI_ALIGNMENT): Define.
6525         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
6526         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
6527         opt is false, only return the psABI mandated alignment increase.
6528         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
6529         (DATA_ABI_ALIGNMENT): ... this.
6530         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
6531         (DATA_ABI_ALIGNMENT): ... this.
6532         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
6533         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
6534         (DATA_ABI_ALIGNMENT): ... this.
6535         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
6536         * doc/tm.texi: Regenerated.
6538 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
6540         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
6541         cmp_code to construct REG_EQUAL note.
6543 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
6545         PR target/57568
6546         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6547         that operands[2] doesn't overlap with operands[0].
6549 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
6550             Jan Hubicka  <jh@suse.cz>
6552         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6553         hack to mark symbols as used.
6555 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
6557         PR rtl-optimization/57559
6558         * lra-constraints.c (process_alt_operands): Don't discourage
6559         memory with known offset for offsetable memory constraint.
6560         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6562 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
6564         * varasm.c (struct oc_local_state): Reorder fields.
6565         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6566         and adjust accordingly.
6567         (output_constructor): Reorder initialization code and adjust call to
6568         output_constructor_bitfield.
6570 2013-06-07  Jan Hubicka  <jh@suse.cz>
6572         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6574 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
6576         * tree-object-size.c (unknown): Make const.
6578 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6580         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6581         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6582         for last alternative in the cpu_facility attribute.
6584 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6586         PR target/56315
6587         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6588         (xordi3): Change operand 2 constraint to arm_xordi_operand.
6589         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6590         * config/arm/constraints.md (Dg): New constraint.
6591         * config/arm/neon.md (xordi3_neon): Remove.
6592         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6593         * config/arm/predicates.md (arm_xordi_operand): New predicate.
6595 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6597         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6598         Clean up alternatives.
6600 2013-06-07  Alan Modra  <amodra@gmail.com>
6602         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6603         va_list_gpr_size.
6605 2013-06-07  Alan Modra  <amodra@gmail.com>
6607         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6609 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6611         * config/arm/constraints.md (Df): New constraint.
6612         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6613         Correct length attribute for last two alternatives.
6615 2013-06-07  Alan Modra  <amodra@gmail.com>
6617         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6618         override user -mfp-in-toc.
6619         (offsettable_ok_by_alignment): Consider just the current access
6620         rather than the whole object, unless BLKmode.  Handle
6621         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6622         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6623         for -mcmodel=medium.
6624         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6625         override user -mfp-in-toc or -msum-in-toc.  Default to
6626         -mno-fp-in-toc for -mcmodel=medium.
6628 2013-06-06  DJ Delorie  <dj@redhat.com>
6630         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6631         TARGET_VALID_POINTER_MODE.
6633 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
6634             Pat Haugen <pthaugen@us.ibm.com>
6635             Peter Bergner <bergner@vnet.ibm.com>
6637         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6638         Document new power8 builtins.
6640         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6641         condition code register, to allow 128-bit logical operations to be
6642         done in the VSX or GPR registers.
6643         (nor<mode>3): Use the canonical form for nor.
6644         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6645         vclz*, and vpopcnt* vector instructions.
6646         (nand<mode>3): Likewise.
6647         (orc<mode>3): Likewise.
6648         (clz<mode>2): LIkewise.
6649         (popcount<mode>2): Likewise.
6651         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6652         that only the GPRs are recognized.
6654         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6655         support for new power8 builtins.
6657         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6658         builtin functions.
6659         (xscvdpspn): Likewise.
6660         (vclz): Likewise.
6661         (vclzb): Likewise.
6662         (vclzh): Likewise.
6663         (vclzw): Likewise.
6664         (vclzd): Likewise.
6665         (vpopcnt): Likewise.
6666         (vpopcntb): Likewise.
6667         (vpopcnth): Likewise.
6668         (vpopcntw): Likewise.
6669         (vpopcntd): Likewise.
6670         (vgbbd): Likewise.
6671         (vmrgew): Likewise.
6672         (vmrgow): Likewise.
6673         (eqv): Likewise.
6674         (eqv_v16qi3): Likewise.
6675         (eqv_v8hi3): Likewise.
6676         (eqv_v4si3): Likewise.
6677         (eqv_v2di3): Likewise.
6678         (eqv_v4sf3): Likewise.
6679         (eqv_v2df3): Likewise.
6680         (nand): Likewise.
6681         (nand_v16qi3): Likewise.
6682         (nand_v8hi3): Likewise.
6683         (nand_v4si3): Likewise.
6684         (nand_v2di3): Likewise.
6685         (nand_v4sf3): Likewise.
6686         (nand_v2df3): Likewise.
6687         (orc): Likewise.
6688         (orc_v16qi3): Likewise.
6689         (orc_v8hi3): Likewise.
6690         (orc_v4si3): Likewise.
6691         (orc_v2di3): Likewise.
6692         (orc_v4sf3): Likewise.
6693         (orc_v2df3): Likewise.
6695         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6696         allow power8 quad mode in 64-bit.
6697         (rs6000_builtin_vectorized_function): Add support to vectorize
6698         ISA 2.07 count leading zeros, population count builtins.
6699         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6700         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6701         (builtin_function_type): Add vgbbd builtin function which takes an
6702         unsigned argument.
6703         (altivec_expand_vec_perm_const): Add support for new power8 merge
6704         instructions.
6706         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
6707         that does not include TImdoe for use with 32-bit.
6708         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
6709         instructions.
6710         (UNSPEC_VSX_CVDPSPN): Likewise.
6711         (vsx_xscvdpspn): Likewise.
6712         (vsx_xscvspdpn): Likewise.
6713         (vsx_xscvdpspn_scalar): Likewise.
6714         (vsx_xscvspdpn_directmove): Likewise.
6715         (vsx_and<mode>3): Split logical operations into 32-bit and
6716         64-bit. Add support to do logical operations on TImode as well as
6717         VSX vector types.  Allow logical operations to be done in either
6718         VSX registers or in general purpose registers in 64-bit mode.  Add
6719         splitters if GPRs were used. For AND, add clobber of CCmode to
6720         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
6721         encoding.
6722         (vsx_and<mode>3_32bit): Likewise.
6723         (vsx_and<mode>3_64bit): Likewise.
6724         (vsx_ior<mode>3): Likewise.
6725         (vsx_ior<mode>3_32bit): Likewise.
6726         (vsx_ior<mode>3_64bit): Likewise.
6727         (vsx_xor<mode>3): Likewise.
6728         (vsx_xor<mode>3_32bit): Likewise.
6729         (vsx_xor<mode>3_64bit): Likewise.
6730         (vsx_one_cmpl<mode>2): Likewise.
6731         (vsx_one_cmpl<mode>2_32bit): Likewise.
6732         (vsx_one_cmpl<mode>2_64bit): Likewise.
6733         (vsx_nor<mode>3): Likewise.
6734         (vsx_nor<mode>3_32bit): Likewise.
6735         (vsx_nor<mode>3_64bit): Likewise.
6736         (vsx_andc<mode>3): Likewise.
6737         (vsx_andc<mode>3_32bit): Likewise.
6738         (vsx_andc<mode>3_64bit): Likewise.
6739         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
6740         and xxlorc instructions.
6741         (vsx_eqv<mode>3_64bit): Likewise.
6742         (vsx_nand<mode>3_32bit): Likewise.
6743         (vsx_nand<mode>3_64bit): Likewise.
6744         (vsx_orc<mode>3_32bit): Likewise.
6745         (vsx_orc<mode>3_64bit): Likewise.
6747         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
6749         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
6750         instruction.
6751         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
6752         (p8_vmrgow): Likewise.
6753         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
6754         GPRs to be split under VSX.
6755         (p8v_clz<mode>2): Add power8 count leading zero support.
6756         (p8v_popcount<mode>2): Add power8 population count support.
6757         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
6758         support.
6760         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
6761         instruction.
6763         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
6764         builtin functions.
6765         (vec_nand): Likewise.
6766         (vec_vclz): Likewise.
6767         (vec_vclzb): Likewise.
6768         (vec_vclzd): Likewise.
6769         (vec_vclzh): Likewise.
6770         (vec_vclzw): Likewise.
6771         (vec_vgbbd): Likewise.
6772         (vec_vmrgew): Likewise.
6773         (vec_vmrgow): Likewise.
6774         (vec_vpopcnt): Likewise.
6775         (vec_vpopcntb): Likewise.
6776         (vec_vpopcntd): Likewise.
6777         (vec_vpopcnth): Likewise.
6778         (vec_vpopcntw): Likewise.
6780 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6782         PR rtl-optimization/57468
6783         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
6784         spilled pseudos.
6786 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6788         PR rtl-optimization/57459
6789         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
6790         type when setting live regs.
6792 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6794         * config/s390/s390.opt (mlra): New option.
6795         * config/s390/s390.c (s390_decompose_address): Check displacement
6796         for all registers for LRA.
6797         (s390_secondary_reload): Don't used secondary reloads for LRA.
6798         (s390_lra_p): New function.
6799         (TARGET_LRA_P): Define.
6800         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
6801         of attribute cpu_facility to zarch for the last alternative.
6802         (*cmpmem_short): Ditto.
6804 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
6806         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6807         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
6808         (arm_expand_prologue): Likewise.
6810 2013-06-06  Teresa Johnson  <tejohnson@google.com>
6812         PR c++/53743
6813         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6814         as this is now done by redirect_edge_and_branch_force.
6815         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6816         barriers, and fix interaction with splitting.
6817         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6818         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6819         reflect changes made in the routine.
6820         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6821         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6822         since this is called in cfglayout mode, and replace partition fixup
6823         with assert as that is now done by force_nonfallthru_and_redirect.
6824         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6825         already be marked with region crossing note.
6826         (insert_section_boundary_note): Make non-static, gate on flag
6827         has_bb_partition, rewrite to also check for multiple partitions.
6828         (rest_of_handle_reorder_blocks): Remove call to
6829         insert_section_boundary_note, now done later during free_cfg.
6830         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6831         * bb-reorder.h (insert_section_boundary_note): Declare.
6832         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6833         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6834         invoke insert_section_boundary_note.
6835         (try_redirect_by_replacing_jump): Remove unnecessary
6836         check for region crossing note.
6837         (fixup_partition_crossing): New function.
6838         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6839         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6840         in non-cfglayout mode.
6841         (force_nonfallthru_and_redirect): Fixup partition boundaries,
6842         remove old code that tried to do this. Emit barrier correctly
6843         when we are in cfglayout mode.
6844         (last_bb_in_partition): New function.
6845         (rtl_split_edge): Correctly fixup partition boundaries.
6846         (commit_one_edge_insertion): Remove old code that tried to
6847         fixup region crossing edge since this is now handled in
6848         split_block, and set up insertion point correctly since
6849         block may now end in a jump.
6850         (verify_hot_cold_block_grouping): Guard against checking when not in
6851         linearized RTL mode.
6852         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6853         notes.
6854         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6855         rtl_verify_flow_info, so not called in cfglayout mode.
6856         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6857         (fixup_reorder_chain): Remove old code that attempted to fixup region
6858         crossing note as this is now handled in force_nonfallthru_and_redirect.
6859         (duplicate_insn_chain): Don't duplicate switch section notes.
6860         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6861         note.
6862         * basic-block.h (emit_barrier_after_bb): Declare.
6864 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6866         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6867         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6868         arm_usatsihi): Adjust alternatives for arm_restrict_it.
6870 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6872         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6873         where appropriate.
6874         * config/arm/ldmstm.md: Regenerate.
6876 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6878         * config/arm/sync.md (atomic_loaddi_1):
6879         Disable predication for arm_restrict_it.
6880         (arm_load_exclusive<mode>): Likewise.
6881         (arm_load_exclusivesi): Likewise.
6882         (arm_load_exclusivedi): Likewise.
6883         (arm_load_acquire_exclusive<mode>): Likewise.
6884         (arm_load_acquire_exclusivesi): Likewise.
6885         (arm_load_acquire_exclusivedi): Likewise.
6886         (arm_store_exclusive<mode>): Likewise.
6887         (arm_store_exclusive<mode>): Likewise.
6888         (arm_store_release_exclusivedi): Likewise.
6889         (arm_store_release_exclusive<mode>): Likewise.
6891 2013-06-06  Richard Biener  <rguenther@suse.de>
6893         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6894         after LTO_null.
6895         (lto_tag_is_tree_code_p): Adjust.
6896         (lto_tag_is_gimple_code_p): Likewise.
6897         (lto_gimple_code_to_tag): Likewise.
6898         (lto_tag_to_gimple_code): Likewise.
6899         (lto_tree_code_to_tag): Likewise.
6900         (lto_tag_to_tree_code): Likewise.
6901         * data-streamer.h (streamer_write_hwi_in_range): Use
6902         uhwi streaming to stream the normalized range.
6903         (streamer_read_hwi_in_range): Likewise.
6905 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6907         * config/arm/arm.md (enabled_for_depr_it): New attribute.
6908         (predicable_short_it): Likewise.
6909         (predicated): Likewise.
6910         (enabled): Handle above.
6911         (define_cond_exec): Set predicated attribute to yes.
6913 2013-06-05  Mike Stump  <mikestump@comcast.net>
6915         * gdbinit.in (__FUNCTION__): Add.
6917 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
6919         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6920         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6922 2013-06-05  Jan Hubicka  <jh@suse.cz>
6924         * varasm.c (mark_decl_referenced): Revert the removal until targets
6925         are fixed.
6927 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6929         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6930         instead of mark_decl_referenced.
6932 2013-06-05  Jan Hubicka  <jh@suse.cz>
6934         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6935         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6936         and symtab_used_from_object_file_p.
6937         (cgraph_make_node_local_1): Clear forced_by_abi.
6938         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6939         * cgraph.h (symtab_node_base): Add forced_by_abi.
6940         (decide_is_variable_needed): Remove.
6941         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6942         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6943         (decide_is_symbol_needed): ... this one; handle symbols in general;
6944         always analyze virtuals; honnor forced_by_abi.
6945         (cgraph_finalize_function): Update.
6946         (varpool_finalize_decl): Update.
6947         (symbol_defined_and_needed): Remove.
6948         (analyze_functions): Update.
6949         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6950         output_refs, input_overwrite_node): Handle forced_by_abi.
6951         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6952         (address_taken_from_non_vtable_p): ... this one.
6953         (comdat_can_be_unshared_p_1): New function.
6954         (cgraph_comdat_can_be_unshared_p): Rename to ...
6955         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6956         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6957         (function_and_variable_visibility): Clear forced_by_abi as needed.
6958         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6959         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6960         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6961         * varpool.c (decide_is_variable_needed): Remove.
6963 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6965         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6966         (arm_option_override): Override arm_restrict_it where appropriate.
6967         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6968         * config/arm/arm.opt (mrestrict-it): New command-line option.
6969         * doc/invoke.texi: Document -mrestrict-it.
6971 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6973         * tsan.c (tsan_atomic_table): Make const.
6975 2013-06-05  Richard Biener  <rguenther@suse.de>
6977         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6978         index associated with the tree we are supposed to replace.
6979         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6980         TREE_ASM_WRITTEN as zero for everything but SSA names.
6982 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6984         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6986 2013-06-04  Jan Hubicka  <jh@suse.cz>
6988         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6989         (input_node, input_varpool_node): Handle correctly external same
6990         body aliases.
6991         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6992         nodes at ltrans stage.
6994 2013-06-04  Jan Hubicka  <jh@suse.cz>
6996         * ipa-inline.c (update_caller_keys): Fix availability test.
6997         (update_callee_keys): Likewise.
6998         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6999         to follow ELF standard.
7001 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
7003         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7004         (mips64r5900el-*-elf*): New configurations.
7005         * config/mips/mips-cpus.def (r5900): New processor.
7006         * config/mips/mips-tables.opt: Regenerate.
7007         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7008         (mips_issue_rate): Handle PROCESSOR_R5900.
7009         (mips_reorg_process_insns): Force reorder mode for the R5900.
7010         * config/mips/mips.h (TARGET_MIPS5900): Define.
7011         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7012         TARGET_MIPS5900.
7013         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7014         TARGET_MIPS5900.
7015         * config/mips/mips.md (processor): Add r5900.
7016         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7018 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7020         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7021         into function to generate MOVI instruction.
7022         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7023         (aarch64_preferred_simd_mode): Turn into wrapper.
7024         (aarch64_output_scalar_simd_mov_immediate): New function.
7025         * config/aarch64/aarch64-protos.h: Add prototype for above.
7027 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7029         * config/aarch64/aarch64.c (simd_immediate_info): Remove
7030         element_char member.
7031         (sizetochar): Return signed char.
7032         (aarch64_simd_valid_immediate): Remove elchar and other
7033         unnecessary variables.
7034         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7035         Calculate element_char as required.
7036         * config/aarch64/aarch64-protos.h: Update and move prototype
7037         for aarch64_output_simd_mov_immediate.
7038         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7039         Update arguments.
7041 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7043         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7044         information completed by aarch64_simd_valid_immediate.
7045         (aarch64_legitimate_constant_p): Update arguments.
7046         (aarch64_simd_valid_immediate): Work with struct rather than many
7047         pointers.
7048         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7049         (aarch64_simd_make_constant): Update arguments.
7050         (aarch64_output_simd_mov_immediate): Work with struct rather than
7051         many pointers.  Output immediate directly rather than as operand.
7052         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7053         Update prototype.
7054         * config/aarch64/constraints.md (Dn): Update arguments.
7056 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7058         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7059         longer static.
7060         (aarch64_simd_immediate_valid_for_move): Remove.
7061         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7062         (aarch64_simd_make_constant): Update call.
7063         (aarch64_output_simd_mov_immediate): Update call.
7064         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7065         Add prototype.
7066         * config/aarch64/constraints.md (Dn): Update call.
7068 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7070         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7071         return type to bool for prototype.
7072         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7073         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7074         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7076 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7078         * config/mips/mips.opt (meva): New.
7079         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7080         (ASM_SPEC): Handle -meva.
7081         * doc/invoke.texi (meva):  Document.
7083 2013-06-04  Alan Modra  <amodra@gmail.com>
7085         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7086         constant output.
7088 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7090         * rtl.def: Add extra fourth optional field to define_cond_exec.
7091         * gensupport.c (process_one_cond_exec): Process attributes from
7092         define_cond_exec.
7093         * doc/md.texi: Document fourth field in define_cond_exec.
7095 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7097         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7098         out the processing order as in store_bit_field_1.
7100 2013-06-04  Jan Hubicka  <jh@suse.cz>
7102         PR middle-end/57500
7103         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7104         non-VAR_DECL node if it does not exist yet.
7106 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
7108         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7109         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7110         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7111         target_cpu_default setting.
7113 2013-06-03  Teresa Johnson  <tejohnson@google.com>
7115         * dumpfile.c (opt_info_switch_p): Change -fopt-info
7116         default to -fopt-info=optimized instead of all.
7117         * doc/invoke.texi: Ditto.
7118         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7119         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7120         (execute_vect_slp): Emit BB vectorization success under
7121         MSG_OPTIMIZED_LOCATIONS.
7122         * tree-vect-slp.c (vect_slp_transform_bb): Change
7123         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7124         * tree-vect-loop.c (vect_transform_loop): Ditto.
7126 2013-06-03  Jason Merrill  <jason@redhat.com>
7128         PR c++/57415
7129         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7130         Use TARGET_EXPR for C++.
7132 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
7134         PR rtl-optimization/57268
7135         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7136         if DEBUG_INSN_P (insn).
7138         Reapply
7139         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7141         PR rtl-optimization/57268
7142         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7143         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7145 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7147         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7148         (ix86_avoid_lea_for_addr): Likewise.
7149         (exact_dependency_1): Likewise.
7150         (ix86_adjust_cost): Likewise.
7151         (swap_top_of_ready_list): Fix formatting and !reload_completed check
7152         removed.
7153         (do_reorder_for_imul): Fix typo, formatting and
7154         !reload_completed check removed.
7155         (ix86_sched_reorder): Fix typo and formatting.
7156         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7157         list.
7159 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
7161         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7163 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
7165         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7166         <STRING_CST>: Likewise.
7167         <VECTOR_CST>: Likewise.
7169 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
7170             Mikael Morin  <mikael@gcc.gnu.org>
7172         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7173         * config.in: Regenerated.
7174         * configure: Regenerated.
7176 2013-06-01  Jan Hubicka  <jh@suse.cz>
7178         PR middle-end/57366
7179         * cgraphunit.c (compile): When weakref is not supported,
7180         set up transparent aliases before final output pass.
7181         * varasm.c (assemble_alias): Do not try to do it here.
7183 2013-06-01  Jan Hubicka  <jh@suse.cz>
7185         PR middle-end/57467
7186         * passes.c (for_per_function): Skip unanalyzed functions.
7188 2013-06-01  Jan Hubicka  <jh@suse.cz>
7190         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7191         (lto_symtab_merge_symbols_1): ... this one.
7192         (lto_symtab_merge_cgraph_nodes): Rename to ...
7193         (lto_symtab_merge_symbols): ... this one; simplify.
7194         * cgraph.c (same_body_aliases_done): Rename to ...
7195         (cpp_implicit_aliases_done): ... this one.
7196         (cgraph_create_function_alias): Update.
7197         (cgraph_same_body_alias): Update.
7198         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7199         (verify_edge_corresponds_to_fndecl): Simplify.
7200         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7201         (cgraph_node): Remove same_body_alias.
7202         (varpool_node): Remove alias_of and extra_name_alias.
7203         (same_body_aliases_done): Rename to ..
7204         (cpp_implicit_aliases_done): ... this one.
7205         (symtab_alias_ultimate_target): Add default parameter.
7206         (symtab_resolve_alias): New function.
7207         (fixup_same_cpp_alias_visibility): Declare.
7208         (cgraph_function_node): Add default parameter.
7209         (cgraph_node_asm_name): Likewise.
7210         (cgraph_function_or_thunk_node): Add default parameter; do
7211         not ICE when it is NULL.
7212         (varpool_variable_node): Likewise.
7213         * tree-emutls.c (create_emultls_var): Update.
7214         (ipa_lower_emutls): Update.
7215         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7216         (cgraph_reset_node): Reset alias info.
7217         (cgraph_finalize_function): Update.
7218         (fixup_same_cpp_alias_visibility): Move to symtab.c.
7219         (analyze_function): Simplify.
7220         (cgraph_process_same_body_aliases): Simplify.
7221         (analyze_functions): Fixup same body aliases.
7222         (handle_alias_pairs): Simplify.
7223         (assemble_thunk): Update.
7224         (assemble_thunks_and_aliases): Update.
7225         (output_weakrefs): Rewrite.
7226         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7227         (lto_output_varpool_node): Likewise.
7228         (compute_ltrans_boundary): Remve assert.
7229         (get_alias_symbol): New functoin.
7230         (input_node): Rewrite alias handling.
7231         (input_varpool_node): Likewise.
7232         * ipa-pure-const.c (propagate_pure_const): Fix formating.
7233         * ipa.c (process_references): Handle weakrefs correctly.
7234         (symtab_remove_unreachable_nodes): Likewise.
7235         * trans-mem.c (get_cg_data): Update.
7236         (ipa_tm_create_version_alias): Update.
7237         (ipa_tm_execute): Update.
7238         * symtab.c (dump_symtab_base): Dump aliases.
7239         (verify_symtab_base): Verify aliases.
7240         (symtab_node_availability): New function.
7241         (symtab_alias_ultimate_target): Simplify.
7242         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7243         handle all the fixup cases.
7244         (symtab_resolve_alias): New function.
7245         * passes.c (ipa_write_summaries): Handle weakrefs.
7246         * varpool.c (varpool_analyze_node): Simplify.
7247         (assemble_aliases): Update.
7248         (varpool_create_variable_alias): Simplify.
7249         (varpool_extra_name_alias): Simplify.
7250         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7251         (lto_symtab_merge_symbols): ... this one.
7253 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
7255         Revert
7256         PR rtl-optimization/57268
7257         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7258         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7260 2013-06-01  Tobias Burnus  <burnus@net-b.de>
7262         Partially reverted:
7263         2013-05-31  Tobias Burnus  <burnus@net-b.de>
7265         PR middle-end/57073
7266         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7267         further up.
7269 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7271         PR rtl-optimization/57268
7272         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7273         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7275 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7277         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7278         unordered comparison operators when -fno-trapping-math is in effect
7279         on the e500.
7280         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7281         and implement unordered comparison operators properly on the e500.
7283 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7285         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7286         for constant scalar integers.
7287         (simplify_relational_operation_1): Likewise.
7289 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
7291         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7292         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
7293         Fix comment.
7295 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7296             Igor Zamyatin  <igor.zamyatin@intel.com>
7298         Silvermont (SLM) architecture performance tuning.
7299         * config/i386/i386.h (enum ix86_tune_indices): Add
7300         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7301         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
7303         * config/i386/i386.c (initial_ix86_tune_features)
7304         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
7305         (ix86_lea_outperforms): Handle Silvermont tuning.
7306         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
7307         call.
7308         (ix86_use_lea_for_mov): Likewise.
7309         (ix86_avoid_lea_for_addr): Likewise.
7310         (ix86_lea_for_add_ok): Likewise.
7311         (exact_dependency_1): New function.
7312         (exact_store_load_dependency): Likewise.
7313         (ix86_adjust_cost): Handle Silvermont tuning.
7314         (do_reoder_for_imul): Likewise.
7315         (swap_top_of_ready_list): New function.
7316         (ix86_sched_reorder): Changed to handle Silvermont tuning.
7318         * config/i386/i386.md (peepholes that split memory operand in fp
7319         converts): New.
7321 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7323         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7324         Remove un-necessary braces.
7326 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7328         * config/aarch64/aarch64.c (aarch64_classify_symbol):
7329         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
7331 2013-05-31  Tobias Burnus  <burnus@net-b.de>
7333         PR middle-end/57073
7334         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
7336 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7338         PR target/56315
7339         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
7340         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
7341         * config/arm/neon.md (iordi3_neon): Remove.
7342         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
7343         * config/arm/predicates.md (imm_for_neon_logic_operand):
7344         Move to earlier in the file.
7345         (neon_logic_op2): Likewise.
7346         (arm_iordi_operand_neon): New predicate.
7348 2013-05-31  Richard Biener  <rguenther@suse.de>
7350         PR tree-optimization/57478
7351         PR tree-optimization/57453
7352         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
7353         are life as well.
7355 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7357         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
7358         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
7360 2013-05-30  Tobias Burnus  <burnus@net-b.de>
7361             Thomas Koenig  <tkoenig@gcc.gnu.org>
7363         PR middle-end/57073
7364         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
7365         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
7367 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
7369         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
7371 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
7373         * target.def (register_usage_leveling_p): New hook.
7374         * targhooks.c (default_register_usage_leveling_p): New.
7375         * targhooks.h (default_register_usage_leveling_p): New prototype.
7376         * lra-assigns.c (register_usage_leveling_p): Use the hook.
7377         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
7378         * doc/tm.texi: Update.
7379         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
7381 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
7383         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
7384         (*insv_reg<mode>): New define_insn.
7386 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
7388         PR rtl-optimization/57439
7389         * postreload.c (move2add_valid_value_p): Check that we have
7390         a zero subreg_regno_offset when accessing the register in
7391         the requested mode.
7393 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7394             Igor Zamyatin  <igor.zamyatin@intel.com>
7396         Silvermont (SLM) architecture pipeline model, tuning and
7397         insn selection.
7398         * config.gcc: Add slm config options and target.
7400         * config/i386/slm.md: New.
7402         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
7404         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
7405         PROCESSOR_SLM.
7406         (ix86_target_macros_internal): Likewise.
7408         * gcc/config/i386/i386.c (slm_cost): New cost.
7409         (m_SLM): New macro flag.
7410         (initial_ix86_tune_features): Set m_SLM.
7411         (x86_accumulate_outgoing_args): Likewise.
7412         (x86_arch_always_fancy_math_387): Likewise.
7413         (processor_target_table): Add slm cost.
7414         (cpu_names): Add slm cpu name.
7415         (x86_option_override_internal): Set SLM ISA.
7416         (ix86_issue_rate): New case PROCESSOR_SLM.
7417         (ia32_multipass_dfa_lookahead): Likewise.
7418         (fold_builtin_cpu): Add slm.
7420         * config/i386/i386.h (TARGET_SLM): New target macro.
7421         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
7422         (processor_type): Add PROCESSOR_SLM.
7424         * config/i386/i386.md (cpu): Add new value "slm".
7425         (slm.md): Include slm.md.
7427 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
7428             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7430         * config/arm/arm-protos.h: Add and update function protos.
7431         * config/arm/arm.c (use_simple_return_p): New added.
7432         (thumb2_expand_return): Check simple_return flag.
7433         * config/arm/arm.md: Add simple_return and conditional simple_return.
7434         * config/arm/iterators.md: Add iterator for return and simple_return.
7436 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7438         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
7439         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
7440         (arm_emit_vfp_multi_reg_pop): Likewise.
7441         (thumb2_emit_ldrd_pop): Likewise.
7442         (arm_expand_epilogue): Add misc REG_CFA notes.
7443         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
7445 2013-05-29  Lawrence Crowl  <crowl@google.com>
7447         * config/arm/t-arm: Update for below.
7449         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
7450         Change type to hash_table.  Update dependent calls and types.
7452         * config/i386/t-cygming: Update for below.
7454         * config/i386/t-interix: Update for below.
7456         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
7457         Change type to hash_table.  Update dependent calls and types.
7458         (i386_find_on_wrapper_list::wrappers): Likewise.
7460         * config/ia64/t-ia64: Update for below.
7462         * config/ia64/ia64.c (bundle_state_table):
7463         Change type to hash_table.  Update dependent calls and types.
7465         * config/mips/mips.c (mips_reorg_process_insns::htab):
7466         Change type to hash_table.  Update dependent calls and types.
7468         * config/sol2.c (solaris_comdat_htab):
7469         Change type to hash_table.  Update dependent calls and types.
7471         * config/t-sol2: Update for above.
7473 2013-05-29  Teresa Johnson  <tejohnson@google.com>
7475         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
7476         functions are not yet marked as defined.
7478 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
7479             Pat Haugen <pthaugen@us.ibm.com>
7480             Peter Bergner <bergner@vnet.ibm.com>
7482         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
7483         instructions.
7484         (VEC_A): Likewise.
7485         (VEC_C): Likewise.
7486         (vrotl<mode>3): Likewise.
7487         (vashl<mode>3): Likewise.
7488         (vlshr<mode>3): Likewise.
7489         (vashr<mode>3): Likewise.
7491         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7492         support for power8 V2DI builtins.
7494         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
7495         power8 V2DI builtins.
7496         (vupkhsw): Likewise.
7497         (vupklsw): Likewise.
7498         (vaddudm): Likewise.
7499         (vminsd): Likewise.
7500         (vmaxsd): Likewise.
7501         (vminud): Likewise.
7502         (vmaxud): Likewise.
7503         (vpkudum): Likewise.
7504         (vpksdss): Likewise.
7505         (vpkudus): Likewise.
7506         (vpksdus): Likewise.
7507         (vrld): Likewise.
7508         (vsld): Likewise.
7509         (vsrd): Likewise.
7510         (vsrad): Likewise.
7511         (vsubudm): Likewise.
7512         (vcmpequd): Likewise.
7513         (vcmpgtsd): Likewise.
7514         (vcmpgtud): Likewise.
7515         (vcmpequd_p): Likewise.
7516         (vcmpgtsd_p): Likewise.
7517         (vcmpgtud_p): Likewise.
7518         (vupkhsw): Likewise.
7519         (vupklsw): Likewise.
7520         (vaddudm): Likewise.
7521         (vmaxsd): Likewise.
7522         (vmaxud): Likewise.
7523         (vminsd): Likewise.
7524         (vminud): Likewise.
7525         (vpksdss): Likewise.
7526         (vpksdus): Likewise.
7527         (vpkudum): Likewise.
7528         (vpkudus): Likewise.
7529         (vrld): Likewise.
7530         (vsld): Likewise.
7531         (vsrad): Likewise.
7532         (vsrd): Likewise.
7533         (vsubudm): Likewise.
7535         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
7536         support for power8 V2DI instructions.
7538         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
7539         power8 V2DI instructions.  Combine pack and unpack insns to use an
7540         iterator for each mode.  Check whether a particular mode supports
7541         Altivec instructions instead of just checking TARGET_ALTIVEC.
7542         (UNSPEC_VPKUWUM): Likewise.
7543         (UNSPEC_VPKSHSS): Likewise.
7544         (UNSPEC_VPKSWSS): Likewise.
7545         (UNSPEC_VPKUHUS): Likewise.
7546         (UNSPEC_VPKSHUS): Likewise.
7547         (UNSPEC_VPKUWUS): Likewise.
7548         (UNSPEC_VPKSWUS): Likewise.
7549         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7550         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7551         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7552         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7553         (UNSPEC_VUPKHSB): Likewise.
7554         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7555         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7556         (UNSPEC_VUPKHSH): Likewise.
7557         (UNSPEC_VUPKLSB): Likewise.
7558         (UNSPEC_VUPKLSH): Likewise.
7559         (VI2): Likewise.
7560         (VI_char): Likewise.
7561         (VI_scalar): Likewise.
7562         (VI_unit): Likewise.
7563         (VP): Likewise.
7564         (VP_small): Likewise.
7565         (VP_small_lc): Likewise.
7566         (VU_char): Likewise.
7567         (add<mode>3): Likewise.
7568         (altivec_vaddcuw): Likewise.
7569         (altivec_vaddu<VI_char>s): Likewise.
7570         (altivec_vadds<VI_char>s): Likewise.
7571         (sub<mode>3): Likewise.
7572         (altivec_vsubcuw): Likewise.
7573         (altivec_vsubu<VI_char>s): Likewise.
7574         (altivec_vsubs<VI_char>s): Likewise.
7575         (altivec_vavgs<VI_char>): Likewise.
7576         (altivec_vcmpbfp): Likewise.
7577         (altivec_eq<mode>): Likewise.
7578         (altivec_gt<mode>): Likewise.
7579         (altivec_gtu<mode>): Likewise.
7580         (umax<mode>3): Likewise.
7581         (smax<mode>3): Likewise.
7582         (umin<mode>3): Likewise.
7583         (smin<mode>3): Likewise.
7584         (altivec_vpkuhum): Likewise.
7585         (altivec_vpkuwum): Likewise.
7586         (altivec_vpkshss): Likewise.
7587         (altivec_vpkswss): Likewise.
7588         (altivec_vpkuhus): Likewise.
7589         (altivec_vpkshus): Likewise.
7590         (altivec_vpkuwus): Likewise.
7591         (altivec_vpkswus): Likewise.
7592         (altivec_vpks<VI_char>ss): Likewise.
7593         (altivec_vpks<VI_char>us): Likewise.
7594         (altivec_vpku<VI_char>us): Likewise.
7595         (altivec_vpku<VI_char>um): Likewise.
7596         (altivec_vrl<VI_char>): Likewise.
7597         (altivec_vsl<VI_char>): Likewise.
7598         (altivec_vsr<VI_char>): Likewise.
7599         (altivec_vsra<VI_char>): Likewise.
7600         (altivec_vsldoi_<mode>): Likewise.
7601         (altivec_vupkhsb): Likewise.
7602         (altivec_vupkhs<VU_char>): Likewise.
7603         (altivec_vupkls<VU_char>): Likewise.
7604         (altivec_vupkhsh): Likewise.
7605         (altivec_vupklsb): Likewise.
7606         (altivec_vupklsh): Likewise.
7607         (altivec_vcmpequ<VI_char>_p): Likewise.
7608         (altivec_vcmpgts<VI_char>_p): Likewise.
7609         (altivec_vcmpgtu<VI_char>_p): Likewise.
7610         (abs<mode>2): Likewise.
7611         (vec_unpacks_hi_v16qi): Likewise.
7612         (vec_unpacks_hi_v8hi): Likewise.
7613         (vec_unpacks_lo_v16qi): Likewise.
7614         (vec_unpacks_hi_<VP_small_lc>): Likewise.
7615         (vec_unpacks_lo_v8hi): Likewise.
7616         (vec_unpacks_lo_<VP_small_lc>): Likewise.
7617         (vec_pack_trunc_v8h): Likewise.
7618         (vec_pack_trunc_v4si): Likewise.
7619         (vec_pack_trunc_<mode>): Likewise.
7621         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7622         V2DI builtins.
7623         (vec_vmaxsd): Likewise.
7624         (vec_vmaxud): Likewise.
7625         (vec_vminsd): Likewise.
7626         (vec_vminud): Likewise.
7627         (vec_vpksdss): Likewise.
7628         (vec_vpksdus): Likewise.
7629         (vec_vpkudum): Likewise.
7630         (vec_vpkudus): Likewise.
7631         (vec_vrld): Likewise.
7632         (vec_vsld): Likewise.
7633         (vec_vsrad): Likewise.
7634         (vec_vsrd): Likewise.
7635         (vec_vsubudm): Likewise.
7636         (vec_vupkhsw): Likewise.
7637         (vec_vupklsw): Likewise.
7639 2013-05-29  Jan Hubicka  <jh@suse.cz>
7641         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7642         flags; reorder rest of fields in more consistent way.
7643         (varpool_node): Remove analyzed, finalized and alias.
7644         (cgraph_ndoe): Likewise.
7645         (symtab_alias_ultimate_target): New function.
7646         (cgraph_function_node): Move offline.
7647         (cgraph_reset_node): Declare.
7648         (cgraph_comdat_can_be_unshared_p): Remove.
7649         (varpool_remove_initializer): Declare.
7650         (varpool_first_defined_variable, varpool_next_defined_variable
7651         cgraph_first_defined_function, cgraph_next_defined_function): Update.
7652         (cgraph_function_with_gimple_body_p): Update.
7653         (varpool_all_refs_explicit_p): Update.
7654         (symtab_alias_target): New function.
7655         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7656         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7657         (cgraph_function_or_thunk_node): Simplify using
7658         symtab_alias_ultimate_target.
7659         (varpool_variable_node): Likewise.
7660         * cgraph.c (cgraph_create_function_alias): Update.
7661         (cgraph_add_thunk): Update.
7662         (cgraph_remove_node): Update.
7663         (dump_cgraph_node): Do not dump removed flags.
7664         (cgraph_function_body_availability): Update.
7665         (cgraph_propagate_frequency): Update.
7666         (verify_cgraph_node): Check sanity of local flag.
7667         (cgraph_function_node): Move here from cgraph.h; revamp for
7668         cgraph_function_or_thunk_node.
7669         * lto-symtab.c (lto_varpool_replace_node): Update.
7670         (lto_symtab_resolve_can_prevail_p): Update.
7671         (lto_symtab_merge_cgraph_nodes): Update.
7672         * ipa-cp.c (determine_versionability, initialize_node_lattices,
7673         propagate_constants_accross_call, devirtualization_time_bonus,
7674         ipcp_propagate_stage): Update.
7675         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7676         * ipa-inline-transform.c (clone_inlined_nodes,
7677         preserve_function_body_p): Update.
7678         * ipa-reference.c (propagate): Update.
7679         (write_node_summary_p): Update.
7680         * toplev.c (wrapup_global_declaration_2): Update.
7681         * cgraphunit.c (cgraph_analyze_function): Rename to ...
7682         (analyze_function) ... this one.
7683         (cgraph_process_new_functions): Update.
7684         (cgraph_reset_node): Export.
7685         (cgraph_finalize_function): Update.
7686         (cgraph_add_new_function): Update.
7687         (process_function_and_variable_attributes): Update.
7688         (varpool_finalize_decl): Update.
7689         (symbol_finalized): Remove.
7690         (symbol_finalized_and_needed): Rename to ...
7691         (symbol_defined_and_needed): ... update.
7692         (cgraph_analyze_functions): Update.
7693         (handle_alias_pairs): Update.
7694         (mark_functions_to_output): Update.
7695         (assemble_thunk): Update.
7696         (output_in_order): Update.
7697         (output_weakrefs): Update.
7698         (finalize_compilation_unit): Update.
7699         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7700         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7701         input_node, input_varpool_node): Update.
7702         * dbxout.c (dbxout_expand_expr): Update.
7703         * cgraphclones.c (cgraph_clone_node): Update.
7704         (cgraph_copy_node_for_versioning): Update.
7705         (cgraph_materialize_clone): Update.
7706         (cgraph_materialize_all_clones): Update.
7707         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
7708         propagate_pure_const, propagate_nothrow): Update.
7709         * lto-streamer-out.c (lto_output, write_symbol): Update.
7710         * ipa-utils.c (ipa_reverse_postorder): Update.
7711         * ipa-inline.c (can_inline_edge_p): Update.
7712         (update_caller_keys, ipa_inline): Update.
7713         * dwarf2out.c (reference_to_unused,
7714         premark_types_used_by_global_vars_helper): Update.
7715         * tree-eh.c (tree_could_trap_p): Update.
7716         * ipa-split.c (consider_split, execute_split_functions): Update.
7717         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
7718         has_addr_references_p): Update; move ahead in file for better
7719         readability.
7720         (process_references): Simplify.
7721         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
7722         bodies are removed.
7723         (cgraph_comdat_can_be_unshared_p): Make static.
7724         (cgraph_externally_visible_p): Update.
7725         (varpool_externally_visible_p): Update.
7726         (function_and_variable_visibility): Update.
7727         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
7728         ipa_tm_mark_force_output_node): Update.
7729         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
7730         estimate_edge_devirt_benefit, inline_generate_summary,
7731         inline_write_summary): Update.
7732         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
7733         * ipa-prop.c (ipa_compute_jump_functions): Update.
7734         (ipa_print_node_params, ipa_prop_read_section,
7735         ipa_update_after_lto_read, read_replacements_section): Update.
7736         * varasm.c (mark_decl_referenced): Update.
7737         (assemble_alias, dump_tm_clone_pairs): Update.
7738         * tree-inline.c (copy_bb): Update.
7739         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
7740         Update.
7741         * symtab.c (dump_symtab_base): Print new flags.
7742         (verify_symtab_base): Verify new flags.
7743         (symtab_alias_ultimate_target): New function.
7744         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
7745         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
7746         Update.
7747         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
7748         Update.
7749         * i386.c (ix86_get_function_versions_dispatcher,
7750         ix86_generate_version_dispatcher_body): Update.
7751         (fold_builtin_cpu): Use varpool_add_new_variable.
7752         * varpool.c (varpool_remove_initializer): Break out from ...
7753         (varpool_remove_node): ... this one.
7754         (dump_varpool_node, varpool_node_for_asm,
7755         cgraph_variable_initializer_availability, varpool_analyze_node,
7756         varpool_assemble_decl, varpool_remove_unreferenced_decls,
7757         varpool_finalize_named_section_flags, varpool_create_variable_alias):
7758         Update.
7760 2013-05-29  Jan Hubicka  <jh@suse.cz>
7762         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
7764 2013-05-29  Easwaran Raman  <eraman@google.com>
7766         PR tree-optimization/57442
7767         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
7768         when control exits the main loop.
7770 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
7772         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
7773         and RX600.
7774         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
7775         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
7776         * rx/t-rx: Add rx100 under multi library matches option for nofpu
7777         option.
7779 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7781         PR tree-optimization/57441
7782         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
7783         Don't limit size of incr_vec to number of candidates.
7785 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
7787         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
7788         and mips16 directories.
7789         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
7790         (MULTILIB_DIRNAMES): Ditto.
7791         (MULTILIB_EXCEPTIONS): Add new exceptions.
7792         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
7793         (MULTILIB_DIRNAMES): Ditto.
7794         (MULTILIB_EXCEPTIONS): Add new exceptions.
7796 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7797             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7799         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
7800         SYMBOL_TINY_ABSOLUTE.
7801         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7802         SYMBOL_TINY_ABSOLUTE.
7803         (aarch64_expand_mov_immediate): Likewise.
7804         (aarch64_classify_symbol): Likewise.
7805         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7806         Permit SYMBOL_TINY_ABSOLUTE.
7807         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7809 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7810             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7812         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7813         Refactor if/switch.  Replace gcc_assert with if.
7815 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7817         * config/i386/i386.c (initial_ix86_tune_features): Enable
7818         FP Reassociation for AMD bdver1 and bdver2.
7820 2013-05-29  Martin Jambor  <mjambor@suse.cz>
7822         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7823         and IMAGPART_EXPR do not occur within other handled_components.
7825 2013-05-29  Richard Biener  <rguenther@suse.de>
7827         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7828         access on whether the use is in the BB we currently try to
7829         vectorize.
7830         (vect_bb_vectorization_profitable_p): Pass the BB we currently
7831         vectorize to vect_bb_slp_scalar_cost.
7833 2013-05-29  Richard Biener  <rguenther@suse.de>
7835         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7836         computing scalar cost offsetted by stmts that are kept live
7837         by scalar uses.
7838         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7839         for computation of scalar cost.
7841 2013-05-28  Steve Ellcey  <sellcey@mips.com>
7843         * config/mips/mips-cpus.def (mips32r2): Change processor type.
7845 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7847         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7848         array notation built-in reduction functions.
7849         * doc/passes.texi (Passes): Added documentation about changes done
7850         for Cilk Plus.
7851         * doc/invoke.texi (C Dialect Options): Added documentation about
7852         the -fcilkplus flag.
7853         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7854         (BUILTINS_DEF): Depend on cilkplus.def.
7855         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
7856         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7857         * cilkplus.def: New file.
7859 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
7861         PR rtl-optimization/57439
7862         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7864 2013-05-28  Easwaran Raman  <eraman@google.com>
7866         PR tree-optimization/57337
7867         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7868         (find_insert_point): Correctly identify the insertion point
7869         when two statements with the same UID is compared.
7871 2013-05-28  Richard Biener  <rguenther@suse.de>
7873         PR tree-optimization/56787
7874         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7875         from the list of data references.
7876         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7877         clobbers.
7878         (vect_analyze_loop_operations): Likewise.
7879         (vect_transform_loop): Remove clobbers.
7881 2013-05-28  Martin Jambor  <mjambor@suse.cz>
7883         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7884         and REALPART_EXPRs have scalar type.
7886 2013-05-28  Richard Biener  <rguenther@suse.de>
7888         PR tree-optimization/57411
7889         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7890         virtual operands.
7891         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7892         virtual operand propagation.
7894 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7896         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7897         destination register for bmasksi_vis.
7898         (vector_init_bshuffle): Likewise.
7899         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7901 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7903         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7904         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7905         mode if the instruction isn't available in the original mode.
7906         * config/sparc/sparc.opt (mfix-ut699): New option.
7907         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7908         (divdf3): Turn into expander.
7909         (divdf3_nofix): New insn.
7910         (divdf3_fix): Likewise.
7911         (divsf3): Disable if -mfix-ut699.
7912         (sqrtdf2): Turn into expander.
7913         (sqrtdf2_nofix): New insn.
7914         (sqrtdf2_fix): Likewise.
7915         (sqrtsf2): Disable if -mfix-ut699.
7917 2013-05-27  Richard Biener  <rguenther@suse.de>
7919         PR middle-end/57412
7920         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7921         block for the new loop.
7923 2013-05-27  Richard Biener  <rguenther@suse.de>
7925         PR tree-optimization/57343
7926         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7927         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7928         (number_of_iterations_cond): Do not build the folded tree.
7930 2013-05-27  Richard Biener  <rguenther@suse.de>
7932         Revert
7933         PR middle-end/57381
7934         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7935         OEP_CONSTANT_ADDRESS_OF retained.
7937         PR tree-optimization/57417
7938         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7939         for unchanged base.
7940         (set_ssa_val_to): Compare addresses using
7941         get_addr_base_and_unit_offset.
7943 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
7945         PR rtl-optimization/56833
7946         * postreload.c (move2add_record_mode): New function.
7947         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7948         (move2add_use_add2_insn): Use move2add_record_sym_value.
7949         (move2add_use_add3_insn): Likewise.
7950         (reload_cse_move2add): Use move2add_valid_value_p and
7951         move2add_record_mode.  Invalidate call-clobbered and REG_INC
7952         affected regs by setting reg_mode to VOIDmode.
7953         (move2add_note_store): Don't pretend the inside of a SUBREG is
7954         the actual destination.  Invalidate single/leading registers by
7955         setting reg_mode to VOIDmode.
7956         Use move2add_record_sym_value, move2add_valid_value_p and
7957         move2add_record_mode.
7959 2013-05-27  Richard Biener  <rguenther@suse.de>
7961         PR tree-optimization/57396
7962         * tree-affine.c (double_int_constant_multiple_p): Properly
7963         return false for val == 0 and div != 0.
7965 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7967         * config/mips/mips.h: Use #elif in preprocessor conditions.
7969 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7971         PR target/53916
7972         * config/mips/constraints.md (kl): New constraint.
7973         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7974         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7975         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7976         from LO for MIPS16.
7977         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7979 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7981         PR target/55777
7982         * config/mips/mips.c (mips_can_inline_p): New function.
7983         (TARGET_CAN_INLINE_P): Define.
7985 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7987         * sched-int.h (ds_t, dw_t): Make unsigned int.
7988         Fix documentation that describes how all the ds_t bits are used.
7989         Reserve the last bit for delayed-branch scheduling.
7990         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7991         (BITS_PER_DEP_WEAK): Fix definition and documentation.
7992         (gen_dep_weak_1): Remove prototype.
7993         * sched-deps.c (get_dep_weak_1): Make static.
7994         * target.def (speculate_insn, needs_block_p, gen_spec_check,
7995         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7996         * doc/tm.texi: Regenerate.
7997         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7999 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
8001         PR debug/56950
8002         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8004 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
8005             Sandra Loosemore  <sandra@codesourcery.com>
8007         * config.gcc (powerpc-*): Allow native for with-cpu.
8009 2013-05-24  Jeff Law  <law@redhat.com>
8011         PR tree-optimization/57124
8012         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8013         conversion feeding a condition if the range has an overflow
8014         if -fstrict-overflow.  Add warnings for when we do make the
8015         transformation.
8017 2013-05-24  Dehao Chen  <dehao@google.com>
8019         * tree-cfg.c (locus_discrim_map): Fix the typo.
8020         (locus_discrim_hasher): Likewise.
8021         (locus_discrim_hasher::hash): Likewise.
8022         (locus_discrim_hasher::equal): Likewise.
8024 2013-05-24  Martin Jambor  <mjambor@suse.cz>
8026         PR tree-optimization/57294
8027         * cgraph.h (ipa_record_stmt_references): Declare.
8028         * cgraphbuild.c (ipa_record_stmt_references): New function.
8029         (build_cgraph_edges): Use ipa_record_stmt_references.
8030         (rebuild_cgraph_edges): Likewise.
8031         (cgraph_rebuild_references): Likewise.
8032         * ipa-prop.c (ipa_modify_call_arguments): Discard references
8033         associated with the old statement and build references from the
8034         newly built statements.
8035         * ipa-ref.c (ipa_remove_stmt_references): New function.
8036         * ipa-ref.h (ipa_remove_stmt_references): Declare.
8038 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
8040         * lra-constraints.c (emit_spill_move): Use smaller mode for
8041         mem-mem moves.
8042         (check_and_process_move): Consider mem-reg moves for secondary
8043         too.
8044         (curr_insn_transform): Don't lose insns emitted before for
8045         secondary memory moves.
8046         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
8047         reg set up in the current insn.
8049 2013-05-24  Dehao Chen  <dehao@google.com>
8051         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8052         hash function.
8053         (locus_descrim_hasher::equal): Likewise.
8054         (build_gimple_cfg): New discriminator assignment algorithm.
8055         (make_edges): Likewise.
8056         (next_discriminator_for_locus): Likewise.
8057         (same_line_p): Likewise.
8058         (assign_discriminators): Likewise.
8059         (make_cond_expr_edges): Likewise.
8060         (make_gimple_switch_edges): Likewise.
8061         (make_goto_expr_edges): Likewise.
8062         (make_gimple_asm_edges): Likewise.
8064 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8066         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8067         X format specifier to only display bottom 16 bits.
8068         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8069         immediate to match for operand 2, since it will be masked.
8071 2013-05-24  Richard Biener  <rguenther@suse.de>
8073         PR tree-optimization/57287
8074         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8075         all SSA names that occur in abnormal PHIs.
8077 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8079         PR tree-ssa/57385
8080         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8081         that index is not negative.
8083 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8085         PR rtl-optimization/55177
8086         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8087         (simplify_byte_swapping_operation): New.
8088         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8089         (simplify_relational_operation_1): Deal with BSWAP.
8091 2013-05-23  Richard Henderson  <rth@redhat.com>
8093         PR target/56742
8094         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8095         (ix86_reorg): Call it.
8097 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8099         PR target/57379
8100         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8101         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8102         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8104 2013-05-23  Christian Bruel  <christian.bruel@st.com>
8106         PR debug/57351
8107         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8109 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8110             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8112         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8113         * config/aarch64/constraints.md (Usa): Remove.
8114         * doc/md.texi (AArch64 Usa): Remove.
8116 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8117             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8119         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8120         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8121         * config/aarch64/predicates.md (aarch64_const_address): Remove.
8122         (aarch64_mov_operand): Use aarch64_mov_operand_p.
8124 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
8126         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8127         instruction (AdvSIMD).
8128         * config/aarch64/aarch64-builtins.c
8129         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8130         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8132 2013-05-23  Martin Jambor  <mjambor@suse.cz>
8134         PR middle-end/57347
8135         * tree.h (contains_bitfld_component_ref_p): Declare.
8136         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8137         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
8138         caller.
8139         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8140         not access a bit-field.  Assert all final offsets are byte-aligned.
8142 2013-05-23  Richard Biener  <rguenther@suse.de>
8144         PR tree-optimization/57380
8145         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8146         least one invariant or re-used load.
8147         * passes.c (init_optimization_passes): Move pass_phiprop before
8148         pass_forwprop.
8150 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
8152         * config/aarch64/aarch64-simd.md
8153         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8155 2013-05-23  Richard Biener  <rguenther@suse.de>
8157         PR middle-end/57381
8158         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8159         OEP_CONSTANT_ADDRESS_OF retained.
8161 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
8163         PR middle-end/57344
8164         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8165         don't lower unit.  Handle unit not being always BITS_PER_WORD.
8167 2013-05-23  Richard Biener  <rguenther@suse.de>
8169         PR rtl-optimization/57341
8170         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8171         instead of true_dependence.
8173 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
8175         * bb-reorder.c (branch_threshold): Make const.
8176         (exec_threshold): Ditto.
8178 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8179             Pat Haugen <pthaugen@us.ibm.com>
8180             Peter Bergner <bergner@vnet.ibm.com>
8182         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8183         documentation for the power8 crypto builtins.
8185         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8187         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8188         macros for defining power8 builtin functions.
8189         (BU_P8V_AV_2): Likewise.
8190         (BU_P8V_AV_P): Likewise.
8191         (BU_P8V_VSX_1): Likewise.
8192         (BU_P8V_OVERLOAD_1): Likewise.
8193         (BU_P8V_OVERLOAD_2): Likewise.
8194         (BU_CRYPTO_1): Likewise.
8195         (BU_CRYPTO_2): Likewise.
8196         (BU_CRYPTO_3): Likewise.
8197         (BU_CRYPTO_OVERLOAD_1): Likewise.
8198         (BU_CRYPTO_OVERLOAD_2): Likewise.
8199         (XSCVSPDP): Fix typo, point to the correct instruction.
8200         (VCIPHER): Add power8 crypto builtins.
8201         (VCIPHERLAST): Likewise.
8202         (VNCIPHER): Likewise.
8203         (VNCIPHERLAST): Likewise.
8204         (VPMSUMB): Likewise.
8205         (VPMSUMH): Likewise.
8206         (VPMSUMW): Likewise.
8207         (VPERMXOR_V2DI): Likewise.
8208         (VPERMXOR_V4SI: Likewise.
8209         (VPERMXOR_V8HI: Likewise.
8210         (VPERMXOR_V16QI: Likewise.
8211         (VSHASIGMAW): Likewise.
8212         (VSHASIGMAD): Likewise.
8213         (VPMSUM): Likewise.
8214         (VPERMXOR): Likewise.
8215         (VSHASIGMA): Likewise.
8217         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8218         __CRYPTO__ if the crypto instructions are available.
8219         (altivec_overloaded_builtins): Add support for overloaded power8
8220         builtins.
8222         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8223         support for power8 crypto builtins.
8224         (builtin_function_type): Likewise.
8225         (altivec_init_builtins): Add support for builtins that take vector
8226         long long (V2DI) arguments.
8228         * config/rs6000/crypto.md: New file, define power8 crypto
8229         instructions.
8231 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8232             Pat Haugen <pthaugen@us.ibm.com>
8233             Peter Bergner <bergner@vnet.ibm.com>
8235         * doc/invoke.texi (Option Summary): Add power8 options.
8236         (RS/6000 and PowerPC Options): Likewise.
8238         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8239         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
8240         wm, wn, wr documentation.
8242         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8243         registers if direct move instructions are enabled.
8244         (wn): New constraint for no registers.
8245         (wq): New constraint for quad word even GPR registers.
8246         (wr): New constraint if 64-bit instructions are enabled.
8247         (wv): New constraint if power8 vector instructions are enabled.
8248         (wQ): New constraint for quad word memory locations.
8250         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8251         constraint for 0..15 for crypto instructions.
8252         (gpc_reg_operand): If VSX allow registers in VSX registers as well
8253         as GPR and floating point registers.
8254         (int_reg_operand): New predicate to match only GPR registers.
8255         (base_reg_operand): New predicate to match base registers.
8256         (quad_int_reg_operand): New predicate to match even GPR registers
8257         for quad memory operations.
8258         (vsx_reg_or_cint_operand): New predicate to allow vector logical
8259         operations in both GPR and VSX registers.
8260         (quad_memory_operand): New predicate for quad memory operations.
8261         (reg_or_indexed_operand): New predicate for direct move support.
8263         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8264         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8265         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8266         (POWERPC_MASKS): Add power8 options.
8267         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8268         various options.
8270         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8271         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8273         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8274         (-mpower8-fusion): New power8 options.
8275         (-mpower8-fusion-sign): Likewise.
8276         (-mpower8-vector): Likewise.
8277         (-mcrypto): Likewise.
8278         (-mdirect-move): Likewise.
8279         (-mquad-memory): Likewise.
8281         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8282         power8.
8283         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8284         registers.
8285         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8286         (rs6000_debug_vector_unit): Add p8_vector.
8287         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8288         definitions.  Also print fusion state.
8289         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8290         (rs6000_builtin_mask_calculate): Add power8 builtin support.
8291         (rs6000_option_override_internal): Add support for power8.
8292         (rs6000_common_init_builtins): Add debugging for skipped builtins
8293         if -mdebug=builtin.
8294         (rs6000_adjust_cost): Add power8 support.
8295         (rs6000_issue_rate): Likewise.
8296         (insn_must_be_first_in_group): Likewise.
8297         (insn_must_be_last_in_group): Likewise.
8298         (force_new_group): Likewise.
8299         (rs6000_register_move_cost): Likewise.
8300         (rs6000_opt_masks): Likewise.
8302         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
8303         power8 capable assembler, default to power7 options.
8304         (TARGET_DIRECT_MOVE): Likewise.
8305         (TARGET_CRYPTO): Likewise.
8306         (TARGET_P8_VECTOR): Likewise.
8307         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
8308         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
8309         (VECTOR_MEM_P8_VECTOR_P): Likewise.
8310         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
8311         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
8312         (TARGET_XSCVDPSPN): Likewise.
8313         (TARGET_XSCVSPDPN): Likewsie.
8314         (TARGET_SYNC_HI_QI): Likewise.
8315         (TARGET_SYNC_TI): Likewise.
8316         (MASK_CRYPTO): Likewise.
8317         (MASK_DIRECT_MOVE): Likewise.
8318         (MASK_P8_FUSION): Likewise.
8319         (MASK_P8_VECTOR): Likewise.
8320         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
8321         temporary used by some of the direct move instructions to get two FP
8322         temporary registers does not force creation of a stack frame.
8323         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
8324         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
8325         that any VSX registers are tieable, even if they are also an
8326         Altivec vector mode.
8327         (r6000_reg_class_enum): Add wm, wr, wv constraints.
8328         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
8329         (RS6000_BTM_CRYPTO): Likewise.
8330         (RS6000_BTM_COMMON): Likewise.
8332         * config/rs6000/rs6000.md (cpu attribute): Add power8.
8333         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
8334         (enum rs6000_vector): Add power8 vector support.
8336 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8338         PR target/19599
8339         PR target/57340
8340         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
8341         (any_sibcall_could_use_r3): this and handle indirect calls.
8342         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
8344 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8346         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
8348 2013-05-22  Richard Biener  <rguenther@suse.de>
8350         PR middle-end/57349
8351         * profile.c (branch_prob): Do not split blocks that are
8352         abnormally receiving from ECF_RETURNS_TWICE functions.
8354 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8356         * recog.c (offsettable_address_addr_space_p): Fix calculation of
8357         address mode.  Move pointer mode initialization to the same place.
8359 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8361         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
8362         while it has any effect.
8364 2013-05-21  Easwaran Raman  <eraman@google.com>
8366         PR tree-optimization/57322
8367         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
8368         UID of the statement added to the BB to be 1.
8370 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
8372         PR tree-optimization/57331
8373         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
8374         of conversion from pointer type to integral type with integer.
8376 2013-05-21  Martin Jambor  <mjambor@suse.cz>
8378         PR lto/57289
8379         * ipa-prop.c (ipa_read_node_info): Process param_used and
8380         controlled_uses in the same order as when writing.
8382 2013-05-21  Magnus Granberg  <baldrick@free.fr>
8384         PR plugins/56754
8385         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
8387 2013-05-21  Richard Biener  <rguenther@suse.de>
8389         PR tree-optimization/57318
8390         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
8391         estimate stmts with side-effects as likely eliminated.
8393 2013-05-21  Richard Biener  <rguenther@suse.de>
8395         PR tree-optimization/57330
8396         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
8397         preserve the call stmts fntype.
8399 2013-05-21  Richard Biener  <rguenther@suse.de>
8401         PR tree-optimization/57303
8402         * tree-ssa-sink.c (statement_sink_location): Improve killing
8403         stmt detection and properly handle self-assignments.
8405 2013-05-21  Christian Bruel  <christian.bruel@st.com>
8407         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
8408         spanning registers. LEAF_REG_REMAP is supported only for contiguous
8409         registers. Set register size out of the PARALLEL loop.
8411 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
8413         PR target/56547
8414         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
8415         (*fmasf4, *fmasf4_media): New insns.
8417 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8419         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
8420         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
8421         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
8422         (mips_idiv_insns): Update the comments to say that the returned
8423         instruction counts are in units of BASE_INSN_LENGTH.
8424         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
8425         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
8426         using 2 rather than 4 as the length of indirect MIPS16 and
8427         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
8428         length of a NOP.  Don't divide MIPS16 lengths by 2.
8429         (mips16_split_long_branches): Assume a branch is long if the
8430         length is greater than 4 rather than 8.
8431         * config/mips/mips.md (length): Give MIPS16 lengths directly,
8432         rather than multiplying them by 2.  Multiply instruction counts
8433         by BASE_INSN_LENGTH rather than 4.
8434         (*jump_mips16, tls_get_tp_mips16_<mode>)
8435         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
8437 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8439         * config/mips/mips.md (extended_mips16): Remove branch case.
8440         (length): Remove duplicated extended_mips16 test.
8442 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8444         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
8446 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
8448         * recog.h (Recog_data): Rename to...
8449         (recog_data_d): ...this.
8450         (recog_data): Update accordingly.
8451         * recog.c (recog_data): Likewise.
8452         * reload.c (save_recog_data): Likewise.
8453         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
8454         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
8456 2013-05-17  Julian Brown  <julian@codesourcery.com>
8458         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
8459         found in a REG_EQUAL note, invalidate it.
8461 2013-05-17  Easwaran Raman  <eraman@google.com>
8463         * tree-ssa-reassoc.c (find_insert_point): New function.
8464         (insert_stmt_after): Likewise.
8465         (get_def_stmt): Likewise.
8466         (ensure_ops_are_available): Likewise.
8467         (not_dominated_by): Likewise.
8468         (rewrite_expr_tree): Do not move statements beyond what is
8469         necessary. Remove call to swap_ops_for_binary_stmt...
8470         (reassociate_bb): ... and move it here.
8471         (build_and_add_sum): Assign UIDs for new statements.
8472         (linearize_expr): Likewise.
8473         (do_reassoc): Renumber gimple statement UIDs.
8475 2013-05-17  Jan Hubicka  <jh@suse.cz>
8477         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
8478         weakrefs.
8479         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8480         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8481         weakrefs.
8482         (output_weakrefs): Update.
8484 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
8485             Martin Jambor  <mjambor@suse.cz>
8487         PR middle-end/57276
8488         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
8489         value that corresponds to the given aggval is found in values vector.
8491 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
8493         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
8494         sse, sse2, sse3, ssse3 and sse4a flags to options.
8496 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
8498         * gengtype-state.c: (s_expr_writer): New class, to handle
8499         prettifying of output layout of s-expressions.
8500         (state_writer): New class, to write out gtype.state.
8501         (state_written_type_count): Move this variable into member data of
8502         state_writer.
8503         (s_expr_writer::s_expr_writer): New code: constructor for new class
8504         (state_writer::state_writer(): ditto
8505         (s_expr_writer::write_new_line): New function
8506         (s_expr_writer::write_any_indent): ditto
8507         (s_expr_writer::begin_s_expr): ditto
8508         (s_expr_writer::end_s_expr): ditto
8509         (write_state_fileloc): convert to method of state_writer...
8510         (state_writer:: write_state_fileloc): ...and use methods of
8511         s_expr_writer to write indentation into the gtype.state output file
8512         to visually represent the hierarchical structure of the list
8513         structures
8514         (write_state_fields): ditto, renaming to...
8515         (state_writer::write_state_fields)
8516         (write_state_a_string): ditto, renaming to...
8517         (state_writer::write_state_a_string)
8518         (write_state_string_option): ditto, renaming to...
8519         (state_writer::write_state_string_option)
8520         (write_state_type_option): ditto, renaming to...
8521         (state_writer::write_state_type_option)
8522         (write_state_nested_option): ditto, renaming to...
8523         (state_writer::write_state_nested_option)
8524         (write_state_option): ditto, renaming to...
8525         (state_writer::write_state_option)
8526         (write_state_options): ditto, renaming to...
8527         (state_writer::write_state_options)
8528         (write_state_lang_bitmap): ditto, renaming to...
8529         (state_writer::write_state_lang_bitmap)
8530         (write_state_version): ditto, renaming to...
8531         (state_writer::write_state_version)
8532         (write_state_scalar_type): ditto, renaming to...
8533         (state_writer::write_state_scalar_type)
8534         (write_state_string_type): ditto, renaming to...
8535         (state_writer::write_state_string_type)
8536         (write_state_undefined_type): ditto, renaming to...
8537         (state_writer::write_state_undefined_type)
8538         (write_state_struct_union_type): ditto, renaming to...
8539         (state_writer::write_state_struct_union_type)
8540         (write_state_struct_type): ditto, renaming to...
8541         (state_writer::write_state_struct_type)
8542         (write_state_user_struct_type): ditto, renaming to...
8543         (state_writer::write_state_user_struct_type)
8544         (write_state_lang_struct_type): ditto, renaming to...
8545         (state_writer::write_state_lang_struct_type)
8546         (write_state_param_struct_type): ditto, renaming to...
8547         (state_writer::write_state_param_struct_type)
8548         (write_state_pointer_type): ditto, renaming to...
8549         (state_writer::write_state_pointer_type)
8550         (write_state_array_type): ditto, renaming to...
8551         (state_writer::write_state_array_type)
8552         (write_state_gc_used): ditto, renaming to...
8553         (state_writer::write_state_gc_used)
8554         (write_state_common_type_content): ditto, renaming to...
8555         (state_writer::write_state_common_type_content)
8556         (write_state_type): ditto, renaming to...
8557         (state_writer::write_state_type)
8558         (write_state_pair_list): ditto, renaming to...
8559         (state_writer::write_state_pair_list)
8560         (write_state_pair): ditto, renaming to...
8561         (state_writer::write_state_pair)
8562         (write_state_typedefs): ditto, renaming to...
8563         (state_writer::write_state_typedefs)
8564         (write_state_structures): ditto, renaming to...
8565         (state_writer::write_state_structures)
8566         (write_state_param_structs): ditto, renaming to...
8567         (state_writer::write_state_param_structs)
8568         (write_state_variables): ditto, renaming to...
8569         (state_writer::write_state_variables)
8570         (write_state_srcdir): ditto, renaming to...
8571         (state_writer::write_state_srcdir)
8572         (write_state_files_list): ditto, renaming to...
8573         (state_writer::write_state_files_list)
8574         (write_state_languages): ditto, renaming to...
8575         (state_writer::write_state_languages)
8576         (write_state): create a state_writer instance and use it when
8577         writing out the state file
8579 2013-05-17  Mike Stump  <mikestump@comcast.net>
8581         PR rtl-optimization/57304
8582         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8583         accessing DF_REF_REAL_LOC.
8585 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8587         PR rtl-optimization/57281
8588         PR rtl-optimization/57300
8589         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8590         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8591         what the other splitter did if the registers are dead.
8593 2013-05-17  Richard Biener  <rguenther@suse.de>
8595         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8596         MEM_REF offsets.
8598 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8600         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8601         linking.
8603 2013-05-17  Marek Polacek  <polacek@redhat.com>
8605         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8606         length when doing non-zero store of storing '\0' to '\0'.
8608 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8610         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8611         vect_external_def oprnd1 with loop_vinfo, try to emit
8612         optional cast, negation and and stmts on the loop preheader
8613         edge instead of into the pattern def seq.
8615         PR tree-optimization/57051
8616         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8617         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8619 2013-05-16  Nick Clifton  <nickc@redhat.com>
8621         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8622         (rl78_is_naked_func): New function.
8623         (rl78_expand_prologue): Skip prologue generation for naked functions.
8624         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8625         * doc/extend.texi (naked): Add RL78 to the list of processors
8626         that supports this attribute.
8628 2013-05-16  Jeff Law  <law@redhat.com>
8630         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8632 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8634         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8635         cache parameters using detect_caches_amd also for CYRIX,
8636         NSC and TM2 signatures.
8638 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8639             Dzianis Kahanovich  <mahatma@eu.by>
8641         PR target/45359
8642         PR target/46396
8643         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8644         VIA/Centaur processors and determine their cache parameters
8645         using detect_caches_amd.
8647 2013-05-16  Teresa Johnson  <tejohnson@google.com>
8649         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8650         (rtl_verify_edges): New function.
8651         (rtl_verify_bb_insns): Ditto.
8652         (rtl_verify_bb_pointers): Ditto.
8653         (rtl_verify_bb_insn_chain): Ditto.
8654         (rtl_verify_fallthru): Ditto.
8655         (rtl_verify_bb_layout): Ditto.
8656         (rtl_verify_flow_info_1): Outline checks into new functions.
8657         (rtl_verify_flow_info): Ditto.
8659 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
8661         * cfghooks.c (copy_bbs): Add update_dominance argument.
8662         * cfghooks.h (copy_bbs): Update prototype.
8663         * tree-cfg.c (gimple_duplicate_sese_region):
8664         Add update_dominance argument.
8665         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8666         * tree-ssa-loop-ch.c (copy_loop_headers): Update
8667         gimple_duplicate_sese_region call.
8668         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8669         Update copy_bbs call.
8670         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8671         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8673 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8675         * tree-vectorizer.h (NUM_PATTERNS): Increment.
8676         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8677         vect_recog_rotate_pattern.
8678         (vect_recog_rotate_pattern): New function.
8680 2013-05-16  Jason Merrill  <jason@redhat.com>
8682         * Makefile.in (LLINKER): New variable.
8683         (mostlyclean): Remove link mutex.
8684         * configure.ac: Handle --enable-link-mutex.
8685         * lock-and-run.sh: New script.
8687 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8689         PR target/19599
8690         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8691         for NULL decl.
8693 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8695         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8697 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
8699         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8700         * config/arm/arm.c (next_consecutive_mem): New function.
8701         (gen_movmem_ldrd_strd): Likewise.
8702         * config/arm/arm.md (movmemqi): Update condition and code.
8703         (unaligned_loaddi, unaligned_storedi): New patterns.
8705 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8707         * config.gcc: Obsolete *-*-solaris2.9*.
8708         * doc/install.texi (Specific, *-*-solaris2*): Document it.
8710 2013-05-16  Richard Biener  <rguenther@suse.de>
8712         * passes.c (init_optimization_passes): Move pass_parallelize_loops
8713         earlier, after GRAPHITE transforms and IV canonicalization.
8715 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8717         * omp-low.c (extract_omp_for_data): For collapsed loops,
8718         if at least one of the loops is known at compile time to
8719         iterate zero times, set count to 0.
8720         (expand_omp_regimplify_p): New function.
8721         (expand_omp_for_generic): For collapsed loops, if at least
8722         one of the loops isn't known to iterate at least once,
8723         add runtime check with setting count to 0.
8724         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
8725         For unsigned types if it isn't known at compile time that
8726         the loop will iterate at least once, add runtime check to bypass
8727         the whole loop if initial condition isn't true.
8729 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
8731         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
8733 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
8735         PR middle-end/57286
8736         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
8737         transformations to avoid an infinite loop.
8739 2013-05-16  Marek Polacek  <polacek@redhat.com>
8741         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
8743 2013-05-15  Leif Ekblad  <leif@rdos.net>
8745         * config/i386/i386.c (ix86_decompose_address): Use
8746         DEFAULT_TLS_SEG_REG to access TLS segment register.
8747         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
8748         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
8749         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
8751 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
8753         PR target/57260
8754         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
8755         sibling calls to functions that would normally be lazily bound,
8756         unless $gp is call-clobbered.
8758 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
8760         * config/i386/i386.c (ix86_option_override_internal): Update
8761         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
8762         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
8763         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
8764         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
8765         of TARGET_3DNOW.
8766         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
8768 2013-05-15  Andreas Schwab  <schwab@suse.de>
8770         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
8771         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
8772         third operand.
8774 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8776         * loop-unroll.c (report_unroll_peel): Check decision before
8777         emitting unroll/peel message.
8779 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8781         * function.h (has_bb_partition): New rtl_data flag.
8782         (bb_reorder_complete): Ditto.
8783         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
8784         instead of flag_reorder_blocks_and_partition.
8785         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
8786         with some enhancements.
8787         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
8788         * bb-reorder.c (connect_traces): Check for has_bb_partition
8789         instead of flag_reorder_blocks_and_partition.
8790         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
8791         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
8792         verify_hot_cold_block_grouping.
8793         (partition_hot_cold_basic_blocks): Set has_bb_partition.
8795 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8797         PR target/19599
8798         * config/arm/predicates.md (call_insn_operand): New predicate.
8799         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
8800         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
8801         if insn is not a tail call.
8802         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8803         registers.
8804         * config/arm/arm.h (enum reg_class): New caller save register class.
8805         (REG_CLASS_NAMES): Likewise.
8806         (REG_CLASS_CONTENTS): Likewise.
8807         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8808         without decls.
8810 2013-05-15  Richard Biener  <rguenther@suse.de>
8812         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8813         of MSG_OPTIMIZED_LOCATIONS.
8814         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8815         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8816         message.
8817         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8818         of MSG_OPTIMIZED_LOCATIONS.
8819         (execute_vect_slp): Likewise.
8820         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8821         (vect_create_cond_for_alias_checks): Likewise.
8822         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8823         (vect_recog_widen_mult_pattern): Likewise.
8824         (vect_recog_widen_sum_pattern): Likewise.
8825         (vect_recog_over_widening_pattern): Likewise.
8826         (vect_recog_widen_shift_pattern): Likewise.
8827         (vect_recog_vector_vector_shift_pattern): Likewise.
8828         (vect_recog_divmod_pattern): Likewise.
8829         (vect_recog_mixed_size_cond_pattern): Likewise.
8830         (vect_recog_bool_pattern): Likewise.
8831         (vect_pattern_recog_1): Likewise.
8833 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8835         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8836         non-functions to builtin_unreachable.
8837         * ipa-inline-transform.c (inline_call): Do not assert estimates were
8838         correct when new direct edges were discovered.
8840 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8842         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8843         header, print symbol order instead of node uid, print more information
8844         about indirect edge targets.
8845         (ipa_make_edge_direct_to_target): Print symbol order instead of node
8846         uids.
8847         (ipa_make_edge_direct_to_target): Likewise.
8848         (remove_described_reference): Likewise.
8849         (propagate_controlled_uses): Likewise.
8850         (ipa_print_node_params): Also print symbol order.
8851         (ipcp_transform_function): Print symbol order instead of node uids.
8852         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8853         (cgraph_get_create_real_symbol_node): Likewise.
8854         * ipa-cp.c (print_lattice): Likewise.
8855         (print_all_lattices): Likewise.
8856         (determine_versionability): Likewise.
8857         (initialize_node_lattices): Likewise.
8858         (estimate_local_effects): Likewise.
8859         (update_profiling_info): Likewise.
8860         (create_specialized_node): Likewise.
8861         (perhaps_add_new_callers): Likewise.
8862         (decide_about_value): Likewise.
8863         (decide_whether_version_node): Likewise.
8864         (identify_dead_nodes): Likewise.
8865         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8866         (dump_inline_summary): Likewise.
8867         (estimate_node_size_and_time): Likewise.
8868         (inline_analyze_function): Likewise.
8869         * ipa-inline.c (report_inline_failed_reason): Likewise.
8870         (want_early_inline_function_p): Likewise.
8871         (edge_badness): Likewise.
8872         (update_edge_key): Likewise.
8873         (inline_small_functions): Likewise.  Add dumping of order to two other
8874         dumps.
8875         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8876         instead of node uids.
8877         (propagate_pure_const): Likewise.
8878         (propagate_pure_const): Likewise.
8879         * ipa-utils.c (dump_cgraph_node_set): Likewise.
8880         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8881         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8882         of node uids.
8883         * tree-pretty-print.c (dump_function_header): Likewise.
8884         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8885         Print symbol order instead of node uids.
8887 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8889         * config/s390/s390.c (s390_register_move_cost): Don't impose the
8890         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8892 2013-05-15  Richard Biener  <rguenther@suse.de>
8894         PR tree-optimization/57275
8895         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8896         return value for fail to do runtime alias checks for gather loads.
8898 2013-05-15  Jan Hubicka  <jh@suse.cz>
8900         PR lto/57038
8901         PR lto/47375
8902         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8903         weakrefs are not external.
8904         (lto_symtab_merge_decls): Fix thinko when dealing with
8905         non-lto_symtab decls.
8906         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8907         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8908         * varpool.c (dump_varpool_node): Dump more flags.
8910 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8912         * config/i386/i386.c (processor_alias_table): Add instruction
8913         FSGSBASE for AMD bdver3 architecture.
8915 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8917         * tree.c (warn_deprecated_use): Print file:line using locus color.
8918         * diagnostic.c (diagnostic_report_current_module): Print file:line
8919         and file:line:column using locus color.
8921 2013-05-14  Mike Stump  <mikestump@comcast.net>
8923         * gdbinit.in: Add __null.
8925 2013-05-14  Mike Stump  <mikestump@comcast.net>
8927         * recog.h: Rename struct recog_data to Recog_data.
8928         * recog.c: Likewise.
8929         * reload.c (can_reload_into): Likewise.
8930         * config/picochip/picochip.c: Likewise.
8932 2013-05-14  Mike Stump  <mikestump@comcast.net>
8934         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8936 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
8938         * resource.h (struct resources): Remove unch_memory member.
8939         (CLEAR_RESOURCE): Don't clear unch_memory.
8940         * resource.c (mark_referenced_resources): Don't set it.
8941         (mark_set_resources): Likewise.
8942         (mark_target_live_regs): Don't clear it.
8943         (init_resource_info): Likewise.
8944         * reorg.c (resource_conflicts_p): Don't compare it.
8945         (redundant_insn): Don't set it.
8947         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8948         Remove prototypes.
8949         * emit-rtl.c (next_label): Remove unused function.
8950         (skip_consecutive_labels, link_cc0_insns): Move to ...
8951         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8952         only place where these functions are used, and make them static.
8954 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8956         * fold-const.c (fold_negate_expr): Handle vectors.
8957         (fold_truth_not_expr): Make it static.
8958         (fold_invert_truthvalue): New static function.
8959         (invert_truthvalue_loc): Handle vectors. Do not call
8960         fold_truth_not_expr directly.
8961         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8962         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8963         (fold_comparison): Handle vector constants.
8964         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8965         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8966         * tree.h (fold_truth_not_expr): Remove declaration.
8968 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8970         * config/aarch64/aarch64-simd.md
8971         (aarch64_vcond_internal<mode>): Rename to...
8972         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8973         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8974         float modes. Clarify all iterator modes.
8975         (vcond<mode><mode>): Use new name for vcond expanders.
8976         (vcond<v_cmp_result><mode>): Likewise.
8977         (vcondu<mode><mode>: Likewise.
8978         * config/aarch64/iterators.md (VDQF_COND): New.
8980 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8982         PR bootstrap/57266
8983         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8984         variable for the shift amount. Check that we shift by non-negative
8985         amounts.
8987 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8989         PR target/42017
8990         * config/arm/arm.h (EPILOGUE_USES): Only return true
8991         for LR_REGNUM after epilogue_completed.
8993 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
8995         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8996         is error_mark_node.
8998 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9000         PR target/57261
9001         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9002         and Solaris 11+/x86 with gld.
9003         * configure: Regenerate.
9005 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9007         * expmed.c (expand_shift_1): Canonicalize rotates by
9008         constant bitsize / 2 to bitsize - 1.
9009         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9010         case ROTATERT>: Likewise.
9012         Revert:
9013         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9015         * config/i386/i386.md (rotateinv): New code attr.
9016         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9017         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9018         roll $31, %eax, etc.
9020 2013-05-14  Richard Biener  <rguenther@suse.de>
9022         PR middle-end/57235
9023         * tree-eh.c (sink_clobbers): Give up for successors with
9024         multiple predecessors and no virtual uses.
9026 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
9028         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9029         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9031 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9033         PR middle-end/57251
9034         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9035         the case when both op0 and op1 have VOIDmode.
9037 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9039         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9040         in multiply-accumulate mode.
9042 2013-05-13  Guozhi Wei  <carrot@google.com>
9044         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9046 2013-05-13  Kai Tietz  <ktietz@redhat.com>
9048         PR target/56975
9049         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9050         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9051         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9052         * config/i386/i386.c (ix86_option_override_internal): Likewise.
9053         (ix86_expand_prologue): Likewise.
9054         (ix86_expand_split_stack_prologue): Likewise.
9055         (legitimate_pic_address_disp_p): Likewise.
9056         (legitimize_pic_address): Likewise.
9057         (legitimize_tls_address): Likewise.
9058         (legitimize_pe_coff_symbol): Likewise.
9059         (output_pic_addr_const): Likewise.
9060         (construct_plt_address): Likewise.
9061         (ix86_expand_call): Likewise.
9062         (x86_output_mi_thunk): Likewise.
9063         (x86_function_profiler): Likewise.
9065 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9067         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9068         similar switch cases.
9069         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9070         (aarch64_simd_mov_to_<mode>low): Delete.
9071         (aarch64_simd_mov_to_<mode>high): Delete.
9072         (move_lo_quad_<mode>): Add w<-r alternative.
9073         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9074         (aarch64_simd_mov_from_*): Update type attribute.
9075         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9076         statement.
9078 2013-05-13  Jan Hubicka  <jh@suse.cz>
9080         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9081         * config/i386/i386.c (ix86_compute_frame_layout,
9082         ix86_expand_epilogue, emit_i387_cw_initialization,
9083         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9084         ix86_local_alignment): Fix use of size/speed predicates.
9086 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9088         PR tree-optimization/45216
9089         PR tree-optimization/57157
9090         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9091         the (-Y) & (B - 1) variant if OP is |.
9092         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9093         return shifted.  Use (-op1) & (prec - 1) as other_amount
9094         instead of prec - op1.
9096 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9098         PR middle-end/42371
9099         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9100         (ipa_constant_data): New type.
9101         (ipa_jump_func): Use ipa_constant_data to hold information about
9102         constant jump functions.
9103         (ipa_get_jf_constant): Adjust to jump function type changes.
9104         (ipa_get_jf_constant_rdesc): New function.
9105         (ipa_param_descriptor): New field controlled_uses.
9106         (ipa_get_controlled_uses): New function.
9107         (ipa_set_controlled_uses): Likewise.
9108         * ipa-ref.h (ipa_find_reference): Declare.
9109         * ipa-prop.c (ipa_cst_ref_desc): New type.
9110         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9111         changes.
9112         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
9113         New parameter cs.  Adjust all callers.
9114         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9115         (remove_described_reference): New function.
9116         (jfunc_rdesc_usable): Likewise.
9117         (try_make_edge_direct_simple_call): Decrement controlled use count,
9118         attempt to remove reference if it hits zero.
9119         (combine_controlled_uses_counters): New function.
9120         (propagate_controlled_uses): Likewise.
9121         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9122         (ipa_edge_duplication_hook): Duplicate reference descriptions.
9123         (ipa_print_node_params): Print described use counter.
9124         (ipa_write_jump_function): Adjust to jump function type changes.
9125         (ipa_read_jump_function): New parameter CS, pass it to
9126         ipa_set_jf_constant.  Adjust caller.
9127         (ipa_write_node_info): Stream controlled use count
9128         (ipa_read_node_info): Likewise.
9129         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9130         asserting.
9131         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9132         count.  Remove cloning-added reference if it reaches zero.
9133         * ipa-ref.c (ipa_find_reference): New function.
9135 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9137         * config/i386/i386.c (processor_target_table): Modified default
9138         alignment values for AMD BD and BT architectures.
9140 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
9142         * tree-vect-generic.c (uniform_vector_p): Move ...
9143         * tree.c (uniform_vector_p): ... here.
9144         * tree.h (uniform_vector_p): Declare it.
9145         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9146         into a scalar.
9148 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9150         PR tree-optimization/57230
9151         * tree-ssa-strlen.c (handle_char_store): Record length for
9152         array store from STRING_CST.
9154         PR tree-optimization/57230
9155         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9156         check.
9158 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
9160         * config/epiphany/epiphany.c (epiphany_init): Check size of
9161         NUM_MODES_FOR_MODE_SWITCHING.
9162         (epiphany_expand_prologue):
9163         Remove CONFIG_REGNUM initial value handling code.
9164         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9165         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9166         (emit_set_fp_mode, epiphany_mode_after): Likewise.
9167         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9168         Don't return 1 for FP_MODE_NONE.
9169         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9170         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9171         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9172         * config/epiphany/epiphany.md (save_config): New pattern.
9174 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
9176         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9178 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
9180         * config/i386/i386.md (memory): Handle sseishft1.
9181         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9182         (*vec_extractv2di_1): Ditto.
9184 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
9186         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9187         saved registers.
9189 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9191         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9192         Add mthumb/march=armv7-a multilib.
9193         Add mthumb/march=armv7-r multilib.
9194         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9196 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
9198         * config/v850/t-rtems: Add more multilibs.
9200 2013-05-10  Richard Biener  <rguenther@suse.de>
9202         PR tree-optimization/57214
9203         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9204         not propagate from SSA names that occur in abnormal PHI nodes.
9206 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
9208         * stor-layout.c (element_precision): New function.
9209         * machmode.h (element_precision): Declare it.
9210         * tree.c (build_minus_one_cst): New function.
9211         (element_precision): Likewise.
9212         * tree.h (build_minus_one_cst): Declare new function.
9213         (element_precision): Likewise.
9214         * fold-const.c (operand_equal_p): Use element_precision.
9215         (fold_binary_loc): Handle vector types.
9216         * convert.c (convert_to_integer): Use element_precision.
9217         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9218         separately.
9220 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
9222         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9223         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9224         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9225         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9226         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9227         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9228         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9229         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9230         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9231         (Uuw8): New constraints.
9232         (Usb4): Move into alphabetical order.
9233         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9234         (sd8_operand, ub8_operand, uw8_operand): New predicates.
9235         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9236         previously unnamed patterns.
9237         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9238         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9239         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9240         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9241         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
9242         of specific lengths.
9244 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9246         * config/i386/i386.md (rotateinv): New code attr.
9247         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9248         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9249         roll $31, %eax, etc.
9251         PR tree-optimization/45216
9252         PR tree-optimization/57157
9253         * tree-ssa-forwprop.c (simplify_rotate): New function.
9254         (ssa_forward_propagate_and_combine): Call it.
9256 2013-05-10  Richard Biener  <rguenther@suse.de>
9258         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9259         disable peeling when we version for aliasing.
9260         (vector_alignment_reachable_p): Honor explicit user alignment.
9261         (vect_supportable_dr_alignment): Likewise.
9262         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9263         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9264         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9265         then peeling to arrange for the cost-model check to come first.
9267 2013-05-10  Alan Modra  <amodra@gmail.com>
9269         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9270         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9271         * configure: Regenerate.
9273 2013-05-10  Alan Modra  <amodra@gmail.com>
9275         PR target/55033
9276         * varasm.c (default_elf_select_section): Move !DECL_P check..
9277         (get_named_section): ..to here before calling get_section_name.
9278         Adjust assertion.
9279         (default_section_type_flags): Add DECL_P check.
9280         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9281         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9283 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
9285         * config/epiphany/epiphany.c (epiphany_expand_prologue):
9286         When using gen_stack_adjust_str with a register offset, add a
9287         REG_FRAME_RELATED_EXPR note.
9289 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
9291         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9292         (*vec_extractv4si_zext_mem): Ditto.
9293         (*vec_extractv2di): Add 0->x and x->x alternatives.
9294         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9295         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9297 2013-05-09  Jason Merrill  <jason@redhat.com>
9299         N3639 C++1y VLA support
9300         * gimplify.c (gimplify_vla_decl): Don't touch an existing
9301         DECL_VALUE_EXPR.
9303         * tree.c (build_constructor_va): New.
9304         * tree.h: Declare it.
9306 2013-05-09  Martin Jambor  <mjambor@suse.cz>
9308         PR lto/57084
9309         * gimple-fold.c (canonicalize_constructor_val): Call
9310         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
9312 2013-05-09  Jan Hubicka  <jh@suse.cz>
9313             Richard Biener  <rguenther@suse.de>
9315         PR lto/54095
9316         * symtab.c (symtab_make_decl_local): Do not add private names.
9318 2013-05-09  Jan Hubicka  <jh@suse.cz>
9320         PR lto/54095
9321         * symtab.c (insert_to_assembler_name_hash): Handle clones.
9322         (unlink_from_assembler_name_hash): Likewise.
9323         (symtab_prevail_in_asm_name_hash, symtab_register_node,
9324         symtab_unregister_node, symtab_initialize_asm_name_hash,
9325         change_decl_assembler_name): Update.
9327 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
9329         * config/aarch64/aarch64.md: New movtf split.
9330         (*movtf_aarch64): Update.
9331         (aarch64_movdi_tilow): Handle TF modes and rename to
9332         aarch64_movdi_<mode>low.
9333         (aarch64_movdi_tihigh): Handle TF modes and rename to
9334         aarch64_movdi_<mode>high
9335         (aarch64_movtihigh_di): Handle TF modes and rename to
9336         aarch64_mov<mode>high_di
9337         (aarch64_movtilow_di): Handle TF modes and rename to
9338         aarch64_mov<mode>low_di
9339         (aarch64_movtilow_tilow): Remove spurious whitespace.
9340         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
9341         splits.
9342         (aarch64_print_operand): Update.
9344 2013-05-09  Alan Modra  <amodra@gmail.com>
9346         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
9347         powerpc64le.
9348         * configure: Regenerate.
9350 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9352         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
9353         splitter preparation statements.
9354         * config/i386/sse.md (*vec_extract* splitters): Ditto.
9355         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
9356         adjust_address_nv.
9358 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9360         * gimple-ssa-strength-reduction.c (count_candidates): Change
9361         return value to int.
9362         (analyze_candidates_and_replace): Change type of length to int.
9364 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9366         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
9367         (*vec_extract<mode>): Use VI12_128 mode iterator.
9368         (*vec_extract<mode>_mem): Ditto.
9369         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
9370         attribute.
9372 2013-05-08  Diego Novillo  <dnovillo@google.com>
9374         PR bootstrap/54659
9376         Revert:
9377         2012-08-17  Diego Novillo  <dnovillo@google.com>
9379         PR bootstrap/54281
9380         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
9381         * config.in: Regenerate.
9382         * configure: Regenerate.
9383         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
9385 2013-05-08  Jan Hubicka  <jh@suse.cz>
9387         PR lto/54095
9388         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
9389         * cgraph.h (symtab_node_base): Add unique_name.
9390         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9391         input_overwrite_node, input_varpool_node): Stream unique_name.
9392         * cgraphclones.c (cgraph_create_virtual_clone,
9393         cgraph_function_versioning): Set unique_name.
9394         * ipa.c (function_and_variable_visibility): Set unique_name.
9396 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9398         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
9399         (alloc_cand_and_find_basis): Restrict conditional candidate
9400         processing to CAND_MULTs.
9402 2013-05-08  Jan Hubicka  <jh@suse.cz>
9404         PR lto/54095
9405         lto-symtab.c (lto_symtab_symbol_p): New function.
9406         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
9407         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
9408         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
9409         Skip static symbols.
9411 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
9413         PR tree-optimization/57200
9414         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
9415         Only call inform if the preceding warning_at returns true.
9417 2013-05-07  Han Shen  <shenhan@google.com>
9419         * cfgexpand.c (record_or_union_type_has_array_p): New function.
9420         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
9421         * common.opt (fstack-protector-strong): New option.
9422         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
9423         * doc/invoke.texi (Optimization Options): Document
9424         "-fstack-protector-strong".
9425         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
9427 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9429         * config/mips/mips.c (mips_machine_reorg2): Return 0.
9431 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
9433         * ira.c (update_equiv_regs): Add insn having equiv memory even if
9434         it is not lhs of the insn.
9435         (setup_reg_equiv): Remove insn having equiv memory which it is not
9436         lhs of the insn.
9437         * lra-constraints.c (process_address): Try to improve generation
9438         code for address base + disp.
9439         (lra_constraints): Make correct the code for checking insn setting
9440         up backward equivalence.  Remove insn only if it is in the init
9441         insn list.
9442         * lra-eliminations.c (update_reg_eliminate): Change return value.
9443         (lra_eliminate): Use the result.
9445 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
9447         * config/i386/sse.md (ssescalarnummask): New mode attribute.
9448         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
9449         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
9450         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
9451         register target operands.
9452         (*vec_extractv8hi_sse2): New pattern.
9453         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
9454         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
9455         (*vec_extract<mode>_mem): New insn and split pattern.
9457 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
9459         * config/arm/arm.c (arm_asan_shadow_offset): New function.
9460         (TARGET_ASAN_SHADOW_OFFSET): Define.
9461         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
9462         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
9464 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9466         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
9467         (incr_vec_index): Return -1 if increment not found.
9468         (create_add_on_incoming_edge): Assert if increment not found.
9469         (record_increment): Limit number of increments recorded.
9470         (all_phi_incrs_profitable): Return false if an increment not found.
9471         (replace_profitable_candidates): Don't process increments that were
9472         not recorded.
9473         (analyze_candidates_and_replace): Limit size of incr_vec.
9475 2013-05-07  Richard Biener  <rguenther@suse.de>
9477         * calls.c (special_function_p): setjmp-like functions are leaf.
9478         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
9479         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
9481 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
9483         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
9484         (aarch64_simd_mov<mode>): New expander.
9485         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
9486         (aarch64_simd_mov_to_<mode>high): Likewise.
9487         (aarch64_simd_mov_from_<mode>low): Likewise.
9488         (aarch64_simd_mov_from_<mode>high): Likewise.
9489         (aarch64_dup_lane<mode>): Update.
9490         (aarch64_dup_lanedi): New instruction pattern.
9491         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
9492         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
9494 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9496         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
9497         (replace_mult_candidate): Remove unnecessary argument; remove
9498         unnecessary parameter from call to introduce_cast_before_cand.
9499         (replace_unconditional_candidate): Remove unnecessary parameter
9500         from call to replace_mult_candidate.
9501         (replace_conditional_candidate): Likewise.
9502         (insert_initializers): Use make_temp_ssa_name.
9503         (introduce_cast_before_cand): Remove unnecessary argument; use
9504         make_temp_ssa_name.
9505         (replace_one_candidate): Remove unnecessary argument; remove
9506         unnecessary parameter from calls to introduce_cast_before_cand.
9507         (replace_profitable_candidates): Remove unnecessary parameters
9508         from calls to replace_one_candidate.
9510 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9512         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9513         phi def as possibly hiding a basis for a CAND_ADD whose operands
9514         have been commuted in the analysis.
9515         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
9517 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
9519         * config/aarch64/aarch64.md
9520         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
9521         shift value between 0-4.
9523 2013-05-07  Richard Biener  <rguenther@suse.de>
9525         * double-int.h (rshift): New overload.
9526         * double-int.c (rshift): New function.
9527         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
9528         (create_reference_ops_from_ref): Remove.
9529         (vn_reference_insert): Use shared ops for constructing the
9530         reference and copy it.
9532 2013-05-07  Richard Biener  <rguenther@suse.de>
9534         PR middle-end/57190
9535         * tree-eh.c (sink_clobbers): Properly propagate
9536         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
9538 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
9540         PR tree-optimization/57149
9541         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9542         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9543         collect_phi_def_edges, execute_late_warn_uninitialized): Use
9544         uninit_undefined_value_p instead of ssa_undefined_value_p.
9546         PR debug/57184
9547         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9548         for modifier == EXPAND_INITIALIZER.
9550 2013-05-07  Anton Blanchard  <anton@samba.org>
9552         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9553         for powerpc64 little endian.
9554         * configure: Regenerate.
9556 2013-05-06  Graham Stott  <grahams@btinternet.com>
9558         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9559         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9560         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9561         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9563 2013-05-06  Graham Stott  <grahams@btinternet.com>
9565         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9566         codes which allow non-lvalues.
9568 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
9570         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9571         components are all 1s.
9572         (integer_minus_onep): New function.
9573         * tree.h (integer_minus_onep): Declare it.
9574         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9575         integer_minus_onep instead of integer_all_onesp.
9577 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9579         PR target/52933
9580         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9581         variations of these patterns.
9583 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9585         * config/i386/i386.md (isa): Add x64_sse4 member.
9586         (enabled): Handle x64_sse4.
9587         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9588         instruction for 64bit SSE4_1 targets.  Update insn attributes.
9589         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9590         instruction for SSE4_1 targets.  Update insn attributes.
9591         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9592         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9593         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9594         const_1 selector.
9595         (*vec_extractv4si): Rename from *sse4_1_pextrd.
9596         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9597         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9599 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9601         PR target/57108
9602         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9604 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
9606         * final.c (do_assembler_dialects): Don't handle curly braces and
9607         vertical bar escaped by % as dialect delimiters.
9608         (output_asm_insn): Print curly braces and vertical bar if escaped
9609         by % and ASSEMBLER_DIALECT defined.
9610         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9611         * doc/tm.texi: Regenerated.
9613 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9615         * config/mips/mips.c: Include tree-pass.h.
9616         (mips_reorg): Split in pre- and post-dbr_schedule parts.
9617         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9618         (pass_mips_machine_reorg2): New machine specific pass.
9619         (insert_pass_mips_machine_reorg2): New pass plugin definition.
9620         (mips_option_override): Register the new pass.
9621         * rtl.h (cleanup_barriers): Remove prototype.
9622         (dbr_schedule): Likewise.
9623         * jump.c (cleanup_barriers): Make static.
9624         * reorg.c (dbr_schedule): Likewise.
9626 2013-05-06  Richard Biener  <rguenther@suse.de>
9628         PR tree-optimization/57185
9629         * tree-parloops.c (add_field_for_reduction): Handle anonymous
9630         SSA names properly.
9632 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9634         PR target/57106
9635         * config/i386/i386.c (add_parameter_dependencies): Add dependence
9636         between "first_arg" and "insn", not "last" and "insn".
9638 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9640         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9641         (find_candidates_in_block): Re-enable slsr_process_phi.
9642         (create_phi_basis): Fix double counting of candidate adjustment.
9644 2013-05-06  Richard Biener  <rguenther@suse.de>
9646         PR middle-end/57147
9647         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9648         the edge is also fallthru, preserve it and just clear the
9649         abnormal flag.
9650         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9651         also complex, preserve that and just clear the fallthru flag.
9652         * tree-inline.c (update_ssa_across_abnormal_edges): Also
9653         update virtual operands.
9655 2013-05-06  Alan Modra  <amodra@gmail.com>
9657         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9658         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9659         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9660         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9661         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9662         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9664 2013-05-06  Alan Modra  <amodra@gmail.com>
9666         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9667         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
9668         (DEFAULT_ASM_ENDIAN): Define.
9669         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9670         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9671         Update -K PIC clause from sysv4.h.
9672         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9673         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9675 2013-05-06  Alan Modra  <amodra@gmail.com>
9677         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9678         twice for little-endian.
9679         (ashrdi3_no_power, ashrdi3): Support little-endian.
9681 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9683         PR target/55303
9684         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9685         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9686         related expanders.
9687         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9688         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9689         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9690         New predicates.
9692 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
9693             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
9695         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9696         * config/pa/pa.opt: Make mbig-switch a no-op.
9697         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9698         (CASE_VECTOR_MODE): Always return SImode.
9699         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9700         for the !TARGET_BIG_SWITCH case.
9701         * config/pa/pa-linux.h: Likewise.
9702         * config/pa/pa-openbsd.h: Likewise.
9703         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9704         * config/pa/pa.md (short_jump): Remove define_insn.
9705         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9706         (casesi0): Remove define_insn.
9707         (type): Remove btable_branch.
9708         (pa_combine_type): Likewise.
9709         (in_nullified_branch_delay): Likewise.
9710         (in_call_delay): Likewise.
9711         (define_delay): Likewise.
9712         (define_insn_reservation "Z3"): Likewise.
9713         (define_insn_reservation "Z4"): Likewise.
9714         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
9715         (pa_adjust_insn_length): Remove adjustment for btable branches.
9716         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
9717         and mno-big-switch
9719 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
9721         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9722         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
9723         Add m->r,x alternatives.
9724         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
9725         splitters using SWI48x mode iterator.
9726         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
9727         TARGET_64BIT.  Add m->x alternative.
9728         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
9729         Add o->x alternative.  Enable for TARGET_SSE.
9730         (sse_storeq): Remove expander.
9731         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
9732         with memory input operand.
9733         (*vec_extractv2di_1 splitter): New.
9734         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
9735         * config/i386/i386.md (ssevecmodelower): New mode attribute.
9737 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9739         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
9740         (INT_LOWPART): Delete.
9741         (extract_MB): Adjust.
9742         (extract_ME): Adjust.
9743         (print_operand): Adjust.
9745 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9747         * config/rs6000/predicates.md (reg_or_add_cint_operand,
9748         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9749         (reg_or_logical_cint_operand, easy_fp_constant,
9750         logical_const_operand): Delete "CONST_DOUBLE" case.
9751         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
9752         "HOST_BITS_PER_WIDE_INT == 64" test.
9753         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
9754         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
9755         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
9756         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
9757         test.
9758         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
9759         CONST_DOUBLE DImode/VOIDmode case.
9760         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
9761         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9762         CONST_DOUBLE VOIDmode case.
9763         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9764         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
9765         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
9766         Delete CONST_DOUBLE case.
9767         (splitters for mov FMOVE64 const_double): Delete
9768         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9769         "HOST_BITS_PER_WIDE_INT >= 64" test.
9770         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
9771         case.
9772         (mov DI const_double): Delete.
9774 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
9776         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
9777         on op shows all bits zero in mode of a lowpart subreg, return zero.
9779 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
9781         PR target/57150
9782         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
9783         to save TFmode registers and DImode to save TImode registers for
9784         caller save operations.
9785         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
9786         mark being partially clobbered since they only use the first
9787         double word.
9789         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
9790         and TDmode only use the upper 64-bits of each VSX register.
9792 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9794         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
9795         (find_candidates_in_block): Disable slsr_process_phi.
9797 2013-05-03  Guozhi Wei  <carrot@google.com>
9799         * coverage.c (coverage_obj_init): Move the construction of gcov
9800         constructor to ...
9801         (build_init_ctor): ... here.
9803 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9805         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9806         (slsr_cand_d): Redefine def_phi.
9807         (stride_status, phi_adjust_status, count_phis_status): New enums.
9808         (find_phi_def): New.
9809         (find_basis_for_base_expr): New.
9810         (find_basis_for_candidate): Handle hidden bases.
9811         (alloc_cand_and_find_basis): Handle phi candidates.
9812         (slsr_process_phi): New.
9813         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9814         (create_mul_imm_cand): Likewise.
9815         (create_add_ssa_cand): Exclude phi base candidates.
9816         (create_add_imm_cand): Likewise.
9817         (slsr_process_cast): Likewise.
9818         (slsr_process_copy): Likewise.
9819         (find_candidates_in_block): Handle phi candidates.
9820         (dump_candidate): Likewise.
9821         (unconditional_cands): Delete.
9822         (unconditional_cands_with_known_stride_p): Delete.
9823         (phi_dependent_cand_p): New.
9824         (cand_increment): Handle phi-dependent candidates.
9825         (replace_dependent): Delete.
9826         (replace_mult_candidate): New.
9827         (replace_unconditional_candidate): New.
9828         (incr_vec_index): Move to avoid forward reference.
9829         (create_add_on_incoming_edge): New.
9830         (create_phi_basis): New.
9831         (replace_dependents): Delete.
9832         (replace_conditional_candidate): New.
9833         (phi_add_costs): New.
9834         (replace_uncond_cands_and_profitable_phis): New.
9835         (record_increment): Handle phi adjustments.
9836         (record_phi_increments): New.
9837         (record_increments): Handle phi adjustments.
9838         (phi_incr_cost): New.
9839         (lowest_cost_path): Handle phis.
9840         (total_savings): Likewise.
9841         (analyze_increments): Likewise.
9842         (ncd_with_phi): New.
9843         (ncd_of_cand_and_phis): New.
9844         (nearest_common_dominator_for_cands): Handle phi increments.
9845         (all_phi_incrs_profitable): New.
9846         (replace_profitable_candidates): Handle phi-dependent candidates.
9847         (analyze_candidates_and_replace): Likewise.
9849 2013-05-03  Teresa Johnson  <tejohnson@google.com>
9851         PR bootstrap/57154
9852         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9853         do not exceed REG_BR_PROB_BASE.
9855 2013-05-03  Jeff Law  <law@redhat.com>
9857         PR tree-optimization/57144
9858         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9859         operand of the condition will bit into the new type when eliminating
9860         a cast feeding a condition.
9862 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
9864         PR rtl-optimization/57130
9865         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9866         of COMPARE as in_code to the recursive call if needed.
9868 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
9870         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9871         (enabled): Handle new members.
9872         * config/i386/sse.md (*vec_concatv2si): Merge from
9873         *vec_concatv2si_sse2 and vec_concatv2si_sse.
9874         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9876 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
9878         PR tree-optimization/57027
9879         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9880         for fnms opportunity, check we got the prerequisite kind
9881         of tree / gimple before using accessor functions.
9883 2013-05-03  Richard Biener  <rguenther@suse.de>
9885         * double-int.h (lshift): New overload without precision
9886         and arith argument.
9887         (operator *=, operator +=, operator -=): Move ...
9888         * double-int.c (operator *=, operator +=, operator -=): ... here
9889         and implement more efficiently.
9890         (mul_double_with_sign): Remove.
9891         (lshift_double): Adjust to take unsinged shift argument, push
9892         dispatching code to callers.
9893         (mul_double_wide_with_sign): Add early out for callers that
9894         are not interested in high parts or overflow.
9895         (lshift): New function.
9896         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9897         dispatch code here.
9898         (lrotate, rrotate): Use logical shifts.
9899         * expr.c (get_inner_reference): Use lshift.
9900         * fixed-value.c (do_fixed_divide): Likewise.
9901         * tree-dfa.c (get_ref_base_and_extent): Likewise.
9902         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9903         (indirect_refs_may_alias_p): Likewise.
9904         (stmt_kills_ref_p_1): Likewise.
9906 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9908         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9910 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9912         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9913         scalar form of FABD instruction.
9915 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
9917         * lra-constraints.c (process_alt_operands): Add checking alt
9918         number to choose the best alternative.
9920 2013-05-02  Richard Biener  <rguenther@suse.de>
9922         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9923         bitmap and its handling.
9924         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9926 2013-05-02  Richard Biener  <rguenther@suse.de>
9928         PR middle-end/57140
9929         * tree-inline.c (copy_loops): Properly handle removed loops.
9930         (copy_cfg_body): Mark destination loops for fixup if source
9931         loops needed fixup.
9933 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9935         PR target/56732
9936         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9937         generating simple_return for naked functions.
9939 2013-05-02  Martin Jambor  <mjambor@suse.cz>
9941         PR middle-end/56988
9942         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9943         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9944         flags match.
9945         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9946         ipa_agg_replacement_value structures.
9947         (known_aggs_to_agg_replacement_list): Likewise.
9948         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9949         (read_agg_replacement_chain): Likewise.
9950         (ipcp_transform_function): Also check that by_ref flags match.
9952 2013-05-02  Richard Biener  <rguenther@suse.de>
9954         * graphds.h (struct graph): Add obstack member.
9955         * graphds.c (new_graph): Initialize obstack and allocate
9956         vertices from it.
9957         (add_edge): Allocate edge from the obstack.
9958         (free_graph): Free the obstack instead of all edges and vertices.
9960 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9962         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9963         divides.
9964         * cfg.c (update_bb_profile_for_threading): Ditto.
9965         * tree-inline.c (copy_bb): Ditto.
9966         (copy_edges_for_bb): Ditto.
9967         (initialize_cfun): Ditto.
9968         (copy_cfg_body): Ditto.
9969         (expand_call_inline): Ditto.
9970         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9971         (estimate_node_size_and_time): Ditto.
9972         (inline_merge_summary): Ditto.
9973         * cgraphclones.c (cgraph_clone_edge): Ditto.
9974         (cgraph_clone_node): Ditto.
9975         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9976         (compute_trg_info): Ditto.
9978 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9980         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9981         S reg when fp attribute set.
9982         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9984 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9986         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9987         New pattern.
9988         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9989         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9990         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9992 2013-05-02  Richard Biener  <rguenther@suse.de>
9994         * tree-scalar-evolution.c (scev_info_hasher): Remove.
9995         (struct instantiate_cache_entry): New type.
9996         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9997         (struct instantiate_cache_type): New type.
9998         (set_instantiated_value, get_instantiated_value): Remove.
9999         (get_instantiated_value_entry): New function.
10000         (instantiate_scev_name): Use the new cache and adjust.
10001         (instantiate_scev_poly): Adjust.
10002         (instantiate_scev_binary): Likewise.
10003         (instantiate_array_ref): Likewise.
10004         (instantiate_scev_convert): Likewise.
10005         (instantiate_scev_not): Likewise.
10006         (instantiate_scev_3): Likewise.
10007         (instantiate_scev_2): Likewise.
10008         (instantiate_scev_r): Likewise.
10009         (instantiate_scev): Likewise.
10010         (resolve_mixers): Likewise.
10012 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
10014         PR target/57091
10015         * lra-constraints.c (best_small_class_operands_num): Remove.
10016         (process_alt_operands): Remove small_class_operands_num.  Take
10017         small classes operands into losers and only if the operand is not
10018         matched.  Modify debugging output.
10019         (curr_insn_transform): Remove best_small_class_operands_num.
10020         Print insn name.
10022 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10024         * config/aarch64/aarch64-builtins.c
10025         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10026         * config/aarch64/aarch64-simd-builtins.def
10027         (reduc_splus_): Add new modes.
10028         (reduc_uplus_): New.
10029         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10030         (reduc_uplus_v4sf): Likewise.
10031         (reduc_splus_v4sf): Likewise.
10032         (aarch64_addv<mode>): Likewise.
10033         (reduc_uplus_<mode>): Likewise.
10034         (reduc_splus_<mode>): Likewise.
10035         (aarch64_addvv2di): Likewise.
10036         (reduc_uplus_v2di): Likewise.
10037         (reduc_splus_v2di): Likewise.
10038         (aarch64_addvv2si): Likewise.
10039         (reduc_uplus_v2si): Likewise.
10040         (reduc_splus_v2si): Likewise.
10041         (reduc_<sur>plus_<mode>): New.
10042         (reduc_<sur>plus_v2di): Likewise.
10043         (reduc_<sur>plus_v2si): Likewise.
10044         (reduc_<sur>plus_v4sf): Likewise.
10045         (aarch64_addpv4sf): Likewise.
10046         * config/aarch64/arm_neon.h
10047         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10048         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10049         add UNSPEC_SADDV, UNSPEC_UADDV.
10050         (SUADDV): New.
10051         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10053 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10055         * config/aarch64/arm_neon.h
10056         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10058 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10060         * config/aarch64/aarch64-builtins
10061         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10063 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10065         * config/aarch64/aarch64-simd-builtins.def
10066         (reduc_smax_): New.
10067         (reduc_smin_): Likewise.
10068         (reduc_umax_): Likewise.
10069         (reduc_umin_): Likewise.
10070         (reduc_smax_nan_): Likewise.
10071         (reduc_smin_nan_): Likewise.
10072         (fmax): Remove.
10073         (fmin): Likewise.
10074         (smax): Update for V2SF, V4SF and V2DF modes.
10075         (smin): Likewise.
10076         (smax_nan): New.
10077         (smin_nan): Likewise.
10078         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10079         (<su><maxmin><mode>3): ...This, refactor.
10080         (s<maxmin><mode>3): New.
10081         (<maxmin_uns><mode>3): Likewise.
10082         (reduc_<maxmin_uns>_<mode>): Refactor.
10083         (reduc_<maxmin_uns>_v4sf): Likewise.
10084         (reduc_<maxmin_uns>_v2si): Likewise.
10085         (aarch64_<fmaxmin><mode>: Remove.
10086         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10087         new builtin names.
10088         (vmin<q>_f<32,64>): Likewise.
10089         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10090         (FMAXMIN): New.
10091         (su): Add mappings for smax, smin, umax, umin.
10092         (maxmin): New.
10093         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10094         (FMAXMIN): Rename as...
10095         (FMAXMIN_UNS): ...This.
10096         (maxminv): Remove.
10097         (fmaxminv): Likewise.
10098         (fmaxmin): Likewise.
10099         (maxmin_uns): New.
10100         (maxmin_uns_op): Likewise.
10102 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10104         * config/aarch64/arm_neon.h
10105         (vac<ge, gt><sd>_f<32, 64>): Rename to...
10106         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10107         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10109 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10111         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10112         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10114 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10116         * config/aarch64/aarch64-simd.md
10117         (vcond<mode>_internal): Handle special cases for constant masks.
10118         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10119         (vcondu<mode><mode>): Likewise.
10120         (vcond<v_cmp_result><mode>): New.
10122 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10124         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10125         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10126         * config/aarch64/aarch64-simd-builtins.def
10127         (cmeq): Update to BUILTIN_VALLDI.
10128         (cmgt): Likewise.
10129         (cmge): Likewise.
10130         (cmle): Likewise.
10131         (cmlt): Likewise.
10132         * config/aarch64/arm_neon.h
10133         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10134         to builtins or C as appropriate.
10136 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10138         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10139         (cmgeu): ...This.
10140         (cmhi): Rename to...
10141         (cmgtu): ...This.
10142         * config/aarch64/aarch64-simd.md
10143         (simd_mode): Add SF.
10144         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10145         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10146         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10147         (cstore<mode>_neg): ...This.
10148         * config/aarch64/iterators.md
10149         (VALLF): new.
10150         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10151         (COMPARISONS): New.
10152         (UCOMPARISONS): Likewise.
10153         (optab): Add missing comparisons.
10154         (n_optab): New.
10155         (cmp_1): Likewise.
10156         (cmp_2): Likewise.
10157         (CMP): Likewise.
10158         (cmp): Remove.
10159         (VCMP_S): Likewise.
10160         (VCMP_U): Likewise.
10161         (V_cmp_result): Add DF, SF modes.
10162         (v_cmp_result): Likewise.
10163         (v): Likewise.
10164         (vmtype): Likewise.
10165         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10167 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
10169         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10170         define_insn to define_insn_and_split.
10171         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10172         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10173         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10174         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10175         (thumb2_negscc): Likewise.
10177 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10179         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10181 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10183         * config/arm/thumb2.md: Remove trailing whitespaces.
10185 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10187         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10188         Use gen_int_mode rather than GEN_INT.
10190 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
10192         * value-prof.c (stream_in_histogram_value): Remove the strayed
10193         debug_gimple_stmt.
10195 2013-04-30  Richard Biener  <rguenther@suse.de>
10197         PR middle-end/57122
10198         * cfghooks.c (split_edge): Properly check for the loop latch edge.
10200 2013-04-30  Richard Biener  <rguenther@suse.de>
10202         PR middle-end/57107
10203         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10205 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
10207         PR rtl-optimization/56957
10208         PR rtl-optimization/57105
10209         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10210         variable.  Use just INSN_UID for determining whether an insn
10211         should be only disconnected from the insn stream.
10212         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10214 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
10216         PR tree-optimization/57104
10217         * tsan.c (instrument_expr): Don't instrument accesses to
10218         DECL_HARD_REGISTER VAR_DECLs.
10220 2013-04-30  Richard Biener  <rguenther@suse.de>
10222         * function.h (loops_for_fn): New inline function.
10223         (set_loops_for_fn): Likewise.
10224         * cfgloop.h (place_new_loop): Add struct function parameter.
10225         (get_loop): Likewise.
10226         (get_loops): Likewise.
10227         (number_of_loops): Likewise.
10228         (fel_next): Adjust.
10229         (fel_init): Likewise.
10230         * cfg.c (get_loop_copy): Adjust.
10231         * cfgloop.c (flow_loops_dump): Likewise.
10232         (record_loop_exits): Likewise.
10233         (verify_loop_structure): Likewise.
10234         * cfgloopanal.c (mark_irreducible_loops): Likewise.
10235         (estimate_reg_pressure_cost): Likewise.
10236         (mark_loop_exit_edges): Likewise.
10237         * cfgloopmanip.c (place_new_loop): Likewise.
10238         (add_loop): Likewise.
10239         (duplicate_loop): Likewise.
10240         * graph.c (draw_cfg_nodes): Likewise.
10241         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10242         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10243         (extract_affine_chrec): Likewise.
10244         (build_scop_iteration_domain): Likewise.
10245         * graphite.c (graphite_initialize): Likewise.
10246         * ira-build.c (create_loop_tree_nodes): Likewise.
10247         (more_one_region_p): Likewise.
10248         (rebuild_regno_allocno_maps): Likewise.
10249         (mark_loops_for_removal): Likewise.
10250         (mark_all_loops_for_removal): Likewise.
10251         (remove_unnecessary_regions): Likewise.
10252         (ira_build): Likewise.
10253         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10254         * loop-init.c (fix_loop_structure): Likewise.
10255         (gate_rtl_move_loop_invariants): Likewise.
10256         (gate_rtl_unswitch): Likewise.
10257         (gate_rtl_unroll_and_peel_loops): Likewise.
10258         (rtl_doloop): Likewise.
10259         * lto-streamer-in.c (input_cfg): Likewise.
10260         * lto-streamer-out.c (output_cfg): Likewise.
10261         * modulo-sched.c (sms_schedule): Likewise.
10262         * predict.c (tree_estimate_probability): Likewise.
10263         (tree_estimate_probability_driver): Likewise.
10264         (estimate_loops): Likewise.
10265         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10266         (move_sese_region_to_fn): Likewise.
10267         (debug_loop_num): Likewise.
10268         * tree-chrec.c (chrec_evaluate): Likewise.
10269         (hide_evolution_in_other_loops_than_loop): Likewise.
10270         (chrec_component_in_loop_num): Likewise.
10271         (reset_evolution_in_loop): Likewise.
10272         (evolution_function_is_invariant_rec_p): Likewise.
10273         * tree-if-conv.c (main_tree_if_conversion): Likewise.
10274         * tree-inline.c (copy_loops): Likewise.
10275         (copy_cfg_body): Likewise.
10276         (tree_function_versioning): Likewise.
10277         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10278         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10279         Likewise.
10280         (add_to_evolution_1): Likewise.
10281         (scev_const_prop): Likewise.
10282         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10283         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10284         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10285         (tree_ssa_lim_initialize): Likewise.
10286         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10287         (verify_loop_closed_ssa): Likewise.
10288         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
10289         (tree_ssa_loop_im): Likewise.
10290         (tree_ssa_loop_unswitch): Likewise.
10291         (tree_vectorize): Likewise.
10292         (check_data_deps): Likewise.
10293         (tree_ssa_loop_ivcanon): Likewise.
10294         (tree_ssa_loop_bounds): Likewise.
10295         (tree_complete_unroll): Likewise.
10296         (tree_complete_unroll_inner): Likewise.
10297         (tree_parallelize_loops): Likewise.
10298         (tree_ssa_loop_prefetch): Likewise.
10299         (tree_ssa_loop_ivopts): Likewise.
10300         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10301         * tree-vectorizer.c (vectorize_loops): Likewise.
10303 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
10305         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
10306         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
10307         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
10308         with EABI_LINK_SPEC.
10310 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10312         PR target/44578
10313         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
10314         alternative.
10316 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
10318         PR target/57097
10319         * lra-constraints.c (process_alt_operands): Discourage a bit more
10320         using memory for pseudos.  Print cost dump for alternatives.
10321         Modify cost values for conflicts with early clobbers.
10322         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
10324 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10326         PR target/57098
10327         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
10329 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10331         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
10332         from/to S register.
10333         (movdi_aarch64): Support LDR/STR from/to D register.
10335 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10337         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
10338         or higher by default.
10340 2013-04-29  Richard Biener  <rguenther@suse.de>
10342         PR middle-end/57075
10343         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
10344         even if not adding abnormal edges for calls that can make
10345         abnormal gotos.
10347 2013-04-29  Richard Biener  <rguenther@suse.de>
10349         PR middle-end/57103
10350         * tree-cfg.c (move_stmt_op): Fix condition under which to update
10351         TREE_BLOCK.
10352         (move_stmt_r): Remove redundant checking.
10354 2013-04-29  Teresa Johnson  <tejohnson@google.com>
10356         PR bootstrap/57077
10357         * basic-block.h (apply_scale): New function.
10358         (apply_probability): Use apply_scale.
10359         * gimple-streamer-in.c (input_bb): Ditto.
10360         * lto-streamer-in.c (input_cfg): Ditto.
10361         * lto-cgraph.c (merge_profile_summaries): Ditto.
10362         * tree-optimize.c (execute_fixup_cfg): Ditto.
10363         * tree-inline.c (copy_bb): Update comment to use apply_scale.
10364         (copy_edges_for_bb): Ditto.
10365         (copy_cfg_body): Ditto.
10367 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10369         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
10370         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
10371         (tail_merge_optimize): Handle current_loops == NULL.
10373 2013-04-26  Jeff Law  <law@redhat.com>
10375         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
10376         (simplify_cond_using_ranges): Generalize code to simplify
10377         COND_EXPRs where one argument is a constant and the other
10378         is an SSA_NAME created by an integral type conversion.
10380 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10382         * config/arm/arm.md (store_minmaxsi): Use only when
10383         optimize_insn_for_size_p.
10385 2013-04-29  Christian Bruel  <christian.bruel@st.com>
10387         PR target/57108
10388         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
10390 2013-04-29  Richard Biener  <rguenther@suse.de>
10392         PR middle-end/57089
10393         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
10394         loop tree make sure to schedule a fixup for the child as well.
10395         (expand_omp_for_generic): Properly add loops.
10396         (expand_omp_for_static_nochunk): Likewise.
10397         (expand_omp_for_static_chunk): Likewise.
10398         (expand_omp_for): For the degenerate case fixup loops.
10399         (expand_omp_sections): Fix default bb placement in loops.
10400         (expand_omp_atomic_pipeline): Properly add loops.
10402 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10404         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
10406 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10408         * tree-ssa-tail-merge.c: Update header comment.
10410 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10412         * config/aarch64/arm_neon.h
10413         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
10414         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
10415         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
10416         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
10417         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
10418         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
10419         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
10420         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
10422 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10424         * config/aarch64/aarch64-simd.md
10425         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
10426         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
10427         fix_trunc, fixuns_trunc.
10428         (ftrunc<VDQF:mode>2): New.
10429         * config/aarch64/iterators.md (optab): Add fix, fixuns.
10430         (fix_trunc_optab): New.
10432 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10434         * config/aarch64/aarch64-builtins.c
10435         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
10436         iceilf, lround, iroundf.
10438 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10440         PR target/54349
10441         * config/i386/i386.h (enum ix86_tune_indices)
10442         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
10443         New, split from X86_TUNE_INTER_UNIT_MOVES.
10444         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
10445         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
10446         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
10447         (TARGET_INTER_UNIT_MOVES): Remove.
10448         * config/i386/i386.c (initial_ix86_tune_features): Update.
10449         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
10450         (ix86_expand_convert_uns_didf_sse): Use
10451         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10452         (ix86_expand_vector_init_one_nonzero): Ditto.
10453         (ix86_expand_vector_init_interleave): Ditto.
10454         (inline_secondary_memory_needed): Return true for moves from SSE class
10455         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
10456         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
10457         * config/i386/constraints.md (Yi, Ym): Depend on
10458         TARGET_INTER_UNIT_MOVES_TO_VEC.
10459         (Yj, Yn): New constraints.
10460         * config/i386/i386.md (*movdi_internal): Change constraints of
10461         operand 1 from Yi to Yj and from Ym to Yn.
10462         (*movsi_internal): Ditto.
10463         (*movdf_internal): Ditto.
10464         (*movsf_internal): Ditto.
10465         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
10466         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10467         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
10468         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
10469         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
10470         * config/i386/sse.md (movdi_to_sse): Ditto.
10471         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
10472         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
10473         TARGET_INTER_UNIT_MOVES.
10474         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
10475         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
10476         instead of TARGET_INTER_UNIT_MOVES.
10477         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
10478         operand 1 from Yi to Yj and from Ym to Yn.
10480 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10482         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
10483         (float_truncate_hi_): Likewise.
10484         (float_extend_lo_): Likewise.
10485         (float_truncate_lo_): Likewise.
10486         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
10487         (aarch64_float_extend_lo_v2df): Likewise.
10488         (vec_unpacks_hi_v4sf): Likewise.
10489         (aarch64_float_truncate_lo_v2sf): Likewise.
10490         (aarch64_float_truncate_hi_v4sf): Likewise.
10491         (vec_pack_trunc_v2df): Likewise.
10492         (vec_pack_trunc_df): Likewise.
10494 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10496         * config/aarch64/aarch64-builtins.c
10497         (aarch64_fold_builtin): Fold float conversions.
10498         * config/aarch64/aarch64-simd-builtins.def
10499         (floatv2si, floatv4si, floatv2di): New.
10500         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
10501         * config/aarch64/aarch64-simd.md
10502         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
10503         * config/aarch64/iterators.md (FLOATUORS): New.
10504         (optab): Add float, floatuns.
10505         (su_optab): Likewise.
10507 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10509         * config/aarch64/aarch64-builtins.c
10510         (aarch64_builtin_vectorized_function): Use new names for
10511         fcvt builtins.
10512         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
10513         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
10514         (fcvtzu): Split as...
10515         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
10516         (fcvtas): Split as...
10517         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
10518         (fcvtau): Split as...
10519         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
10520         (fcvtps): Split as...
10521         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
10522         (fcvtpu): Split as...
10523         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
10524         (fcvtms): Split as...
10525         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
10526         (fcvtmu): Split as...
10527         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
10528         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
10529         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
10530         (lfrintnusf, lfrintnudf): Likewise.
10531         * config/aarch64/aarch64-simd.md
10532         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
10533         define_insn.
10534         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
10535         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
10536         (fcvt_pattern): Likewise.
10538 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10540         * config/aarch64/aarch64-simd.md
10541         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10542         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10544 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10546         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10547         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10548         (vrnd<a,m,n,p>_f32): Implement using builtins.
10549         (vrnd<i,x><q>_f<32, 64>): New.
10551 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10553         * config/aarch64/aarch64-builtins.c
10554         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10555         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10556         (frintz): Rename to...
10557         (btrunc): ...this.
10558         (frintp): Rename to...
10559         (ceil): ...this.
10560         (frintm): Rename to...
10561         (floor): ...this.
10562         (frinti): Rename to...
10563         (nearbyint): ...this.
10564         (frintx): Rename to...
10565         (rint): ...this.
10566         (frinta): Rename to...
10567         (round): ...this.
10568         * config/aarch64/aarch64-simd.md
10569         (aarch64_frint<frint_suffix><mode>): Delete.
10570         (<frint_pattern><mode>2): Convert to insn.
10571         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10572         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10573         (frint_pattern): Likewise.
10574         (frint_suffix): Likewise.
10576 2013-04-29  Richard Biener  <rguenther@suse.de>
10578         PR tree-optimization/57081
10579         * loop-init.c: Include tree-flow.h.
10580         (loop_optimizer_finalize): Free number of iteration estimates.
10581         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10583 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
10585         PR tree-optimization/57083
10586         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10587         non-singleton shift count range, zero extend low_bound for uns case.
10589         * config/i386/predicates.md (general_vector_operand): New predicate.
10590         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10591         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10592         if they aren't nonimmediate operands.  If their original values
10593         satisfy const_vector_equal_evenodd_p, don't shift them.
10594         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10595         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
10596         if not nonimmediate_operand.
10597         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10598         instead of register_operand.
10599         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10601 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
10603         * stor-layout.c (finalize_size_functions): Allocate a structure and
10604         reset cfun before dumping the functions.
10606 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
10608         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10610         PR target/56866
10611         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10612         use xop_pmacsdqh if uns_p.
10613         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10614         the immediate rotate count.
10616 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
10618         * rtl.h (struct rtx_def): Add comment for field jump.
10619         (LRA_SUBREG_P): New macro.
10620         * recog.c (register_operand): Check LRA_SUBREG_P.
10621         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10622         stack frame.
10623         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10624         (lra_final_code_change): Skip subreg change for operators.
10625         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10626         if there are no operand changes.
10627         * lra-constraints.c (curr_insn_set): New.
10628         (match_reload): Set LRA_SUBREG_P.
10629         (emit_spill_move): Ditto.
10630         (check_and_process_move): Use curr_insn_set. Process only single
10631         set insns.  Don't initialize sec_mem_p and change_p.
10632         (simplify_operand_subreg): Use LRA_SUBREG_P.
10633         (reg_in_class_p): New function.
10634         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10635         of #ifdef.  Add code to remove cycling.
10636         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10637         non-null disp.  Reload inner instead of disp when base and index
10638         are null.  Try to put lo_sum into register.
10639         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10640         (check_and_process_move): Move code for move cost check to
10641         simple_move_p.  Remove equiv_substitution.
10642         (simple_move_p): New function.
10643         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10644         curr_insn_set.  Call check_and_process_move only for single set
10645         insns.  Use the new function.  Move call of check_and_process_move
10646         after operand equiv substitution and address process.
10648 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10650         PR go/57045
10651         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10652         with nonlocal goto receivers or returns twice calls, ignore
10653         unininitialized values from abnormal edges to nl goto receiver
10654         or returns twice call.
10656 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10658         PR tree-optimization/57051
10659         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10660         and VEC_RSHIFT_EXPR if shift count is a multiple of element
10661         bitsize.
10663 2013-04-26  Richard Biener  <rguenther@suse.de>
10665         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10666         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
10667         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10668         (fixup_loop_arrays_after_move): New function.
10669         (move_sese_region_to_fn): Properly outline the loop tree parts
10670         of the SESE region.
10672 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
10674         * config/i386/i386.md (type, unit): Fix long lines.
10676 2013-04-26  Richard Biener  <rguenther@suse.de>
10678         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10679         (lto-streamer-out.o): Likewise.
10680         * cfgloop.c (init_loops_structure): Export, add struct function
10681         argument and adjust.
10682         (flow_loops_find): Adjust.
10683         * cfgloop.h (enum loop_estimation): Add EST_LAST.
10684         (init_loops_structure): Declare.
10685         * lto-streamer-in.c: Include cfgloop.h.
10686         (input_cfg): Input the loop tree.
10687         * lto-streamer-out.c: Include cfgloop.h.
10688         (output_cfg): Output the loop tree.
10689         (output_struct_function_base): Do not drop PROP_loops.
10691 2013-03-26  Richard Biener  <rguenther@suse.de>
10693         * tree-cfg.c (execute_build_cfg): Build the loop tree.
10694         (pass_build_cfg): Provide PROP_loops.
10695         (move_sese_region_to_fn): Remove loops that are outlined into fn
10696         for now.
10697         * tree-inline.c: Include cfgloop.h.
10698         (initialize_cfun): Do not drop PROP_loops.
10699         (copy_loops): New function.
10700         (copy_cfg_body): Copy loop structure.
10701         (tree_function_versioning): Initialize destination loop tree.
10702         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10703         (pass_parallelize_loops): Do IL verification.
10704         * loop-init.c (loop_optimizer_init): Fixup loops if required.
10705         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10706         the CFG make sure we fixup loops as well.
10707         * tree-ssa-tail-merge.c: Include cfgloop.h.
10708         (replace_block_by): When merging loop latches mark loops for fixup.
10709         * lto-streamer-out.c (output_struct_function_base): Drop
10710         PROP_loops for now.
10711         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
10712         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
10713         * ipa-split.c: Include cfgloop.h.
10714         (split_function): Add the new return block to the loop tree root.
10715         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
10716         whether we have removed the forwarder block.
10717         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
10718         * cfgloop.h (place_new_loop): Declare.
10719         * cfgloopmanip.c (place_new_loop): Export.
10720         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
10721         (tree-switch-conversion.o): Likewise.
10722         (tree-complex.o): Likewise.
10723         (tree-inline.o): Likewise.
10724         (tree-ssa-tailmerge.o): Likewise.
10725         (ipa-split.o): Likewise.
10726         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
10727         (tree-ssa-copy.o): Likewise.
10728         * tree-switch-conversion.c: Include cfgloop.h
10729         (process_switch): If we emit a bit-test cascade, schedule loops
10730         for fixup.
10731         * tree-complex.c: Include cfgloop.h.
10732         (expand_complex_div_wide): Properly add new basic-blocks to loops.
10733         * asan.c: Include cfgloop.h.
10734         (create_cond_insert_point): Properly add new basic-blocks to
10735         loops, schedule loop fixup.
10736         * cfgloop.c (verify_loop_structure): Check that looks are not
10737         marked for fixup.
10738         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
10739         to loops.
10740         (expand_omp_for_generic): Likewise.
10741         (expand_omp_sections): Likewise.
10742         (expand_omp_atomic_pipeline): Schedule loops for fixup.
10743         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
10744         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
10745         is initialized, not when loops are present.
10746         * tree-parloops.c (parallelize_loops): Remove checking here.
10747         * passes.c (init_optimization_passes): Schedule a copy-propagation
10748         pass before complete unrolling of inner loops.
10750 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10752         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
10753         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
10754         (colorize_init): Add argument to _WIN32 version.
10755         * toplev.c: Include diagnostic-color.h.
10756         (process_options): Default to -fdiagnostics-color=auto if
10757         GCC_COLORS env var is in the environment.
10758         * common.opt (fdiagnostics-color=): Add Var and Init.
10759         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
10760         env var is in the environment, the default is auto rather than never.
10762         * diagnostic.h (file_name_as_prefix): Add context argument.
10763         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
10764         the string as locus.
10765         * langhooks.c (lhd_print_error_function): Adjust caller.
10767 2013-04-25  Lawrence Crowl  <crowl@google.com>
10769         * var-tracking.c (shared_hash_def::htab):
10770         Change type to hash_table.  Update dependent calls and types.
10772 2013-04-25  Lawrence Crowl  <crowl@google.com>
10774         * Makefile.in: Update as needed below.
10776         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
10777         Move declaration to after the type's method definitons.
10779         * attribs.c (htab_t scoped_attributes::attribute_hash):
10780         Change type to hash_table.  Update dependent calls and types.
10782         * bitmap.c (htab_t bitmap_desc_hash):
10783         Change type to hash_table.  Update dependent calls and types.
10785         * cselib.c (htab_t cselib_hash_table):
10786         Change type to hash_table.  Update dependent calls and types.
10788         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
10789         (hash_string_slot_node): Move implementation into lto-streamer.h
10790         struct string_slot_hasher.
10791         (eq_string_slot_node): Likewise.
10793         * data-streamer-out.c: Update output_block::string_hash_table
10794         dependent calls and types.
10796         * dwarf2cfi.c (htab_t trace_index):
10797         Change type to hash_table.  Update dependent calls and types.
10799         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
10800         Change type to hash_table.  Update dependent calls and types.
10801         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10802         (htab_t optimize_external_refs::map): Likewise.
10803         (htab_t output_comp_unit::extern_map): Likewise.
10804         (htab_t output_comdat_type_unit::extern_map): Likewise.
10805         (htab_t output_macinfo::macinfo_htab): Likewise.
10806         (htab_t optimize_location_lists::htab): Likewise.
10807         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10809         * except.c (htab_t ehspec_hash_type):
10810         Change type to hash_table.  Update dependent calls and types.
10811         (assign_filter_values::ttypes): Likewise.
10812         (assign_filter_values::ehspec): Likewise.
10813         (sjlj_assign_call_site_values::ar_hash): Likewise.
10814         (convert_to_eh_region_ranges::ar_hash): Likewise.
10816         * gcse.c (htab_t pre_ldst_table):
10817         Change type to hash_table.  Update dependent calls and types.
10819         * ggc-common.c (htab_t saving_htab):
10820         Change type to hash_table.  Update dependent calls and types.
10821         (htab_t loc_hash): Likewise.
10822         (htab_t ptr_hash): Likewise.
10823         (call_count): Rename ggc_call_count.
10824         (call_alloc): Rename ggc_call_alloc.
10825         (loc_descriptor): Rename make_loc_descriptor.
10826         (add_statistics): Rename ggc_add_statistics.
10828         * ggc-common.c (saving_htab):
10829         Change type to hash_table.  Update dependent calls and types.
10831         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10832         (push_gimplify_context): Likewise.
10833         (pop_gimplify_context): Likewise.
10834         (struct gimple_temp_hash_elt): Added.
10835         (struct gimplify_hasher): Likewise.
10836         (struct gimplify_ctx.temp_htab):
10837         Change type to hash_table.  Update dependent calls and types.
10839         * gimple-fold.c: Include gimplify-ctx.h.
10841         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10842         Change type to hash_table.  Update dependent calls and types.
10843         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10844         avoid potential global name collision.
10846         * gimplify.c: Include gimplify-ctx.h.
10847         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10848         (htab_t gimplify_ctx::temp_htab):
10849         Update dependent calls and types for new type hash_table.
10850         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10851         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10853         * gimplify-ctx.h: New.
10854         (struct gimple_temp_hash_elt): Move from gimplify.c.
10855         (class gimplify_hasher): New.
10856         (struct gimplify_ctx): Move from gimple.h.
10857         (htab_t gimplify_ctx::temp_htab):
10858         Change type to hash_table.  Update dependent calls and types.
10860         * graphite-clast-to-gimple.c: Include graphite-htab.h.
10861         (htab_t ivs_params::newivs_index):
10862         Change type to hash_table.  Update dependent calls and types.
10863         (htab_t ivs_params::params_index): Likewise.
10864         (htab_t print_generated_program::params_index): Likewise.
10865         (htab_t gloog::newivs_index): Likewise.
10866         (htab_t gloog::params_index): Likewise.
10868         * graphite.c: Include graphite-htab.h.
10869         4htab_t graphite_transform_loops::bb_pbb_mapping):
10870         Change type to hash_table.  Update dependent calls and types.
10872         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10873         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10874         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10876         * graphite-dependences.c: Include graphite-htab.h.
10877         (loop_is_parallel_p): Change hash table type of parameter.
10879         * graphite-htab.h: New.
10880         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10881         (extern find_pbb_via_hash): Move from graphite-poly.h.
10882         (extern loop_is_parallel_p): Move from graphite-poly.h.
10883         (extern get_loop_body_pbbs): Move from graphite-poly.h.
10885         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10886         (extern loop_is_parallel_p): Move to graphite-htab.h.
10887         (extern get_loop_body_pbbs): Move to graphite-htab.h.
10889         * haifa-sched.c (htab_t delay_htab):
10890         Change type to hash_table.  Update dependent calls and types.
10891         (htab_t delay_htab_i2): Likewise.
10893         * ira-color.c (htab_t allocno_hard_regs_htab):
10894         Change type to hash_table.  Update dependent calls and types.
10896         * ira-costs.c (htab_t cost_classes_htab):
10897         Change type to hash_table.  Update dependent calls and types.
10899         * loop-invariant.c (htab_t merge_identical_invariants::eq):
10900         Change type to hash_table.  Update dependent calls and types.
10902         * loop-iv.c (htab_t bivs):
10903         Change type to hash_table.  Update dependent calls and types.
10905         * loop-unroll.c (htab_t opt_info::insns_to_split):
10906         Change type to hash_table.  Update dependent calls and types.
10907         (htab_t opt_info::insns_with_var_to_expand): Likewise.
10909         * lto-streamer.h (struct string_slot): Move from data-streamer.h
10910         (struct string_slot_hasher): New.
10911         (htab_t output_block::string_hash_table):
10912         Change type to hash_table.  Update dependent calls and types.
10914         * lto-streamer-in.c (freeing_string_slot_hasher): New.
10915         (htab_t file_name_hash_table):
10916         Change type to hash_table.  Update dependent calls and types.
10918         * lto-streamer-out.c: Update output_block::string_hash_table dependent
10919         calls and types.
10921         * lto-streamer.c (htab_t tree_htab):
10922         Change type to hash_table.  Update dependent calls and types.
10924         * omp-low.c: Include gimplify-ctx.h.
10926         * passes.c (htab_t name_to_pass_map):
10927         Change type to hash_table.  Update dependent calls and types.
10928         (pass_traverse): Rename to passes_pass_traverse.
10930         * plugin.c (htab_t event_tab):
10931         Change type to hash_table.  Update dependent calls and types.
10933         * postreload-gcse.c (htab_t expr_table):
10934         Change type to hash_table.  Update dependent calls and types.
10935         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10937         * sese.c (debug_rename_map_1): Make extern.
10938         (htab_t copy_bb_and_scalar_dependences::rename_map):
10939         Change type to hash_table.  Update dependent calls and types.
10941         * sese.h (extern debug_rename_map): Move to .c file.
10943         * store-motion.c (htab_t store_motion_mems_table):
10944         Change type to hash_table.  Update dependent calls and types.
10946         * trans-mem.c (htab_t tm_new_mem_hash):
10947         Change type to hash_table.  Update dependent calls and types.
10949         * tree-browser.c (htab_t TB_up_ht):
10950         Change type to hash_table.  Update dependent calls and types.
10952         * tree-cfg.c (htab_t discriminator_per_locus):
10953         Change type to hash_table.  Update dependent calls and types.
10955         * tree-complex.c: Include tree-hasher.h
10956         (htab_t complex_variable_components):
10957         Change type to hash_table.  Update dependent calls and types.
10959         * tree-eh.c (htab_t finally_tree):
10960         Change type to hash_table.  Update dependent calls and types.
10962         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10963         struct int_tree_hasher.
10964         (extern int_tree_map_eq): Likewise.
10965         (uid_decl_map_hash): Removed.
10966         (extern decl_tree_map_eq): Likewise.
10968         * tree-hasher.h: New.
10969         (struct int_tree_hasher): New.
10970         (typedef int_tree_htab_type): New.
10972         * tree-inline.c: Include gimplify-ctx.h.
10974         * tree-mudflap.c: Include gimplify-ctx.h.
10976         * tree-parloops.c: Include tree-hasher.h.
10977         (htab_t eliminate_local_variables_stmt::decl_address):
10978         Change type to hash_table.  Update dependent calls and types.
10979         (htab_t separate_decls_in_region::decl_copies): Likewise.
10981         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10982         Change type to hash_table.  Update dependent calls and types.
10984         * tree-sra.c (candidates):
10985         Change type to hash_table.  Update dependent calls and types.
10987         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10988         in tree-flow.h.
10989         (int_tree_map_hash): Likewise.
10991         * tree-ssa-dom.c (htab_t avail_exprs):
10992         Change type to hash_table.  Update dependent calls and types.
10994         * tree-ssa-live.c (var_map_base_init::tree_to_index):
10995         Change type to hash_table.  Update dependent calls and types.
10997         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10998         Change type to hash_table.  Update dependent calls and types.
11000         * tree-ssa-phiopt.c (seen_ssa_names):
11001         Change type to hash_table.  Update dependent calls and types.
11003         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11004         Change type to hash_table.  Update dependent calls and types.
11006         * tree-ssa-uncprop.c (equiv):
11007         Change type to hash_table.  Update dependent calls and types.
11009 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
11011         PR rtl-optimization/57003
11012         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11013         call note_stores with kill_clobbered_value callback again after
11014         killing regs_invalidated_by_call.
11016 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11018         * config/aarch64/aarch64-simd.md
11019         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11020         (aarch64_simd_bsl<mode>): Likewise.
11021         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11023 2013-04-25  Marek Polacek  <polacek@redhat.com>
11025         PR tree-optimization/57066
11026         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11028 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
11030         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11032 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11034         * config/aarch64/aarch64-builtins.c
11035         (aarch64_fold_builtin): New.
11036         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11037         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11038         * config/aarch64/aarch64-simd-builtins.def (abs): New.
11039         * config/aarch64/arm_neon.h
11040         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11042 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11043             Tejas Belagod  <tejas.belagod@arm.com>
11045         * config/aarch64/aarch64-builtins.c
11046         (aarch64_gimple_fold_builtin): New.
11047         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11048         * config/aarch64/aarch64-simd-builtins.def (addv): New.
11049         * config/aarch64/aarch64-simd.md (addpv4sf): New.
11050         (addvv4sf): Update.
11051         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11053 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11055         * config/aarch64/aarch64.md
11056         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11058 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11060         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11061         (*ngcsi_uxtw): New pattern.
11063 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11064             Julian Brown  <julian@codesourcery.com>
11066         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11067         (TB_DREG): Add T_V4HF.
11068         (v4hf_UP): New macro.
11069         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11070         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11071         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11072         built-ins.
11073         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11074         (arm_vector_mode_supported_p): Handle V4HF.
11075         (arm_mangle_map): Handle V4HFmode.
11076         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11077         * config/arm/arm_neon_builtins.def: Add entries for
11078         vcvtv4hfv4sf, vcvtv4sfv4hf.
11079         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11080         (neon_vcvtv4hfv4sf): Likewise.
11081         * config/arm/neon-gen.ml: Handle half-precision floating point
11082         features.
11083         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11084         * config/arm/arm_neon.h: Regenerate.
11085         * config/arm/neon.ml (type elts): Add F16.
11086         (type vectype): Add T_float16x4, T_floatHF.
11087         (type vecmode): Add V4HF.
11088         (type features): Add Requires_FP_bit feature.
11089         (elt_width): Handle F16.
11090         (elt_class): Likewise.
11091         (elt_of_class_width): Likewise.
11092         (mode_of_elt): Refactor.
11093         (type_for_elt): Handle F16, fix error messages.
11094         (vectype_size): Handle T_float16x4.
11095         (vcvt_sh): New function.
11096         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11097         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11098         (string_of_mode): Handle V4HF.
11099         * doc/arm-neon-intrinsics.texi: Regenerate.
11101 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11103         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11104         format specifier in 'X' case.
11106 2013-04-25  Alan Modra  <amodra@gmail.com>
11108         PR target/57052
11109         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11110         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11111         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11112         Repeat for many other rotate/shift and mask patterns using subregs.
11113         Name lshiftrt insns.
11114         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11115         on WORDS_BIG_ENDIAN.
11117 2013-04-25  Alan Modra  <amodra@gmail.com>
11119         * config.gcc: Support little-endian powerpc-linux targets.
11120         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11121         (LINK_OS_LINUX_SPEC): Define.
11122         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11123         Preserve MASK_LITTLE_ENDIAN.
11124         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11125         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11126         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11127         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11128         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11129         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11130         Correct fp word order for little-endian.  Don't shift toc entries
11131         smaller than a word for little-endian.
11132         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11133         (bswapdi2 splits): Correct low-part subreg for little-endian.
11134         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11135         low/high where such is correct only for be.
11136         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11137         little-endian for -mcall-aixdesc.
11139 2013-04-25  Alan Modra  <amodra@gmail.com>
11141         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11142         replace_equiv_address_nv.
11144 2013-04-25  Alan Modra  <amodra@gmail.com>
11146         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11148 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11150         Revert:
11151         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11152         * rtl.h (struct rtx_def): ...
11154 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11156         PR rtl-optimizations/57046
11157         * lra-constraints (split_reg): Set up lra_risky_transformations_p
11158         for multi-reg splits.
11160 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
11162         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11164 2013-04-24  Sterling Augustine  <saugustine@google.com>
11166         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11167         (comp_dir_string, debug_str_dwo_section): New.
11168         (DEBUG_STR_DWO_SECTION): Rename to ...
11169         (DEBUG_DWO_STR_SECTION): ... this.
11170         (DEBUG_NORM_STR_SECTION): Delete.
11171         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11172         (DEBUG_STR_DWO_SECTION_FLAGS): New.
11173         (find_AT_string): Move most logic to ...
11174         (find_AT_string_in_table): ... here.  New.
11175         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11176         add_skeleton_AT_string.  Delete logic.
11177         (output_skeleton_debug_sections): Remove call to
11178         add_top_level_skeleton_die_attrs.
11179         (add_comp_dir_attribute): Move logic to comp_dir_string.
11180         (dwarf2out_init): Initialize debug_str_dwo_section.
11181         (output_indirect_string): Call find_string_form.
11182         (output_indirect_strings): Rewrite.
11183         (prune_unused_types): Empty skeleton_debug_str_hash.
11184         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11185         (dwarf2out_finish):  Call output_indirect_strings.
11187 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11189         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11191 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11193         * rtl.h (struct rtx_def): Add comment for field jump.
11194         (LRA_SUBREG_P): New macro.
11195         * recog.c (register_operand): Check LRA_SUBREG_P.
11196         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11197         stack frame.
11198         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11199         (lra_final_code_change): Skip subreg change for operators.
11200         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11201         if there are no operand changes.
11202         * lra-constraints.c (curr_insn_set): New.
11203         (match_reload): Set LRA_SUBREG_P.
11204         (emit_spill_move): Ditto.
11205         (check_and_process_move): Use curr_insn_set. Process only single
11206         set insns.  Don't initialize sec_mem_p and change_p.
11207         (simplify_operand_subreg): Use LRA_SUBREG_P.
11208         (reg_in_class_p): New function.
11209         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11210         of #ifdef.  Add code to remove cycling.
11211         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11212         non-null disp.  Reload inner instead of disp when base and index
11213         are null.  Try to put lo_sum into register.
11214         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11215         (check_and_process_move): Move code for move cost check to
11216         simple_move_p.  Remove equiv_substitution.
11217         (simple_move_p): New function.
11218         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11219         curr_insn_set.  Call check_and_process_move only for single set
11220         insns.  Use the new function.  Move call of check_and_process_move
11221         after operand equiv substitution and address process.
11223 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
11225         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11226         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11227         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11229 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11231         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11233 2013-04-24  Marek Polacek  <polacek@redhat.com>
11235         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11236         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11237         (select_loops_exit_conditions): Likewise.
11238         (number_of_iterations_for_all_loops): Likewise.
11239         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11240         (scev_analysis): Likewise.
11242 2013-04-02  Catherine Moore  <clm@codesourcery.com>
11243             Chao-ying Fu <fu@mips.com>
11245         * config/mips/micromips.md (jraddiusp): New pattern.
11246         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11247         instruction if possible.
11249 2013-04-24  Alan Modra  <amodra@gmail.com>
11251         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11253 2013-04-24  Julian Brown  <julian@codesourcery.com>
11254             Chung-Lin Tang  <cltang@codesourcery.com>
11256         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11257         dependency behavior in enumeration type DIE generation. Add TODO note
11258         to comments about future DW_FORM_sdata/udata re-work of related code.
11260 2013-04-23  Lawrence Crowl  <crowl@google.com>
11262         * Makefile.in: Update as needed below.
11264         * hash-table.h (class hash_table):
11265         Correct many methods with parameter types compare_type to the correct
11266         value_type.  (Correct code was unlikely to notice the change.)
11267         (hash_table::elements_with_deleted) New.
11268         (class hashtable::iterator): New.
11269         (hashtable::begin()): New.
11270         (hashtable::end()): New.
11271         (FOR_EACH_HASH_TABLE_ELEMENT): New.
11273         * statistics.c (statistics_hashes):
11274         Change type to hash_table.  Update dependent calls and types.
11276         * tree-into-ssa.c (var_infos):
11277         Change type to hash_table.  Update dependent calls and types.
11279         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11280         Change type to hash_table.  Update dependent calls and types.
11282         * tree-ssa-loop-im.c (struct mem_ref.refs):
11283         Change type to hash_table.  Update dependent calls and types.
11285         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11286         Change type to hash_table.  Update dependent calls and types.
11288         * tree-ssa-sccvn.c (vn_tables_s::nary):
11289         Change type to hash_table.  Update dependent calls and types.
11290         (vn_tables_s::phis): Likewise.
11291         (vn_tables_s::references): Likewise.
11293         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11294         (vn_reference_eq): Update parameter and return types.
11296         * tree-ssa-structalias.c (pointer_equiv_class_table):
11297         Change type to hash_table.  Update dependent calls and types.
11298         (location_equiv_class_table): Likewise.
11300         * tree-vect-data-refs.c: Consequential changes for making
11301         peeling a hash_table.
11303         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
11304         (destroy_loop_vec_info): Dependent hash_table update.
11306         * tree-vectorizer.h (peeling_htab):
11307         Change type to hash_table.  Update dependent calls and types.
11309 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
11311         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
11312         to check the register content is equal or not.
11313         * lra-constraints.c (match_reload): Use lra_assign_reg_val
11314         to assign register content record.
11315         * lra-eliminations.c (update_reg_eliminate): Use
11316         lra_update_reg_val_offset to update register content offset.
11317         * lra-int.h (struct lra_reg): Add offset member.
11318         (lra_reg_val_equal_p): New static inline function.
11319         (lra_update_reg_val_offset): New static inline function.
11320         (lra_assign_reg_val): New static inline function.
11321         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
11322         to assign register content record.
11323         (initialize_lra_reg_info_element): Initial offset to zero.
11325 2013-04-23  Catherine Moore  <clm@codesourcery.com>
11327         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
11328         operands.  Record compression.
11330 2013-04-23  Xinliang David Li  <davidxl@google.com>
11332         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
11334 2013-04-23  Richard Biener  <rguenther@suse.de>
11336         PR middle-end/57036
11337         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
11338         parameter, only add abnormal goto edges from the copied body
11339         if the call could perform abnormal gotos.
11340         (copy_cfg_body): Adjust.
11342 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
11344         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
11346 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
11348         * coretypes.h (gimple_stmt_iterator): Add struct to make
11349         compatible with C.
11351 2013-04-23  Richard Biener  <rguenther@suse.de>
11353         PR tree-optimization/57026
11354         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
11355         from SSA names occuring in abnormal PHI nodes.
11357 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11359         * lto/lto.c (print_lto_report_1): Fix LTO report names.
11361 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11363         * lto/lto.c (print_lto_report_1): Declare early.
11364         (read_cgraph_and_symbols): Call print_lto_report_1 early.
11366 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11368         * common.opt (-flto-report-wpa): Add.
11369         * doc/invoke.texi (-flto-report-wpa): Add.
11370         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
11371         (lto_main): dito.
11373 2013-04-22  Xinliang David Li  <davidxl@google.com>
11375         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
11376         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
11377         * Makefile.in: New dependency
11379             David Daney <ddaney.cavm@gmail.com>
11381         * configure.ac (gcc_cv_as_micromips_support):  Use the
11382         --fatal-warnings option.
11383         * configure: Regenerate.
11385 2013-04-22  Marek Polacek  <polacek@redhat.com>
11387         PR sanitizer/56990
11388         * tsan.c (instrument_expr): Don't instrument expression
11389         in case its size is zero.
11391 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
11393         PR target/57032
11394         Revert:
11395         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
11397         * config/alpha/alpha.c (TARGET_LRA_P): New define.
11399 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11401         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
11402         (gimple_stmt_iterator): New typedef.
11403         * gimple.h (gimple_stmt_iterator): Rename to...
11404         (gimple_stmt_iterator_d): ... This.
11405         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
11406         trees be valid for GIMPLE and GENERIC.
11407         (TARGET_GIMPLE_FOLD_BUILTIN): New.
11408         * gimple-fold.c (gimple_fold_call): Call target hook
11409         gimple_fold_builtin.
11410         * hooks.c (hook_bool_gsiptr_false): New.
11411         * hooks.h (hook_bool_gsiptr_false): New.
11412         * target.def (fold_stmt): New.
11413         * doc/tm.texi: Regenerate.
11415 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
11417         PR target/57018
11418         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
11419         a set sp if no stack realignment.
11421 2013-04-22  Nick Clifton  <nickc@redhat.com>
11423         * config.gcc (tilegx-linux): Extend extra_objs rather than
11424         overwriting it.
11425         (tilepro-linux): Likewise.
11427 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11429         * config/aarch64/aarch64-builtins.c
11430         (CF): Remove.
11431         (CF0, CF1, CF2, CF3, CF4, CF10): New.
11432         (VAR<1-12>): Add MAP parameter.
11433         (BUILTIN_*): Likewise.
11434         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
11435         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
11436         (aarch64_ushl_n<mode>): Likewise.
11437         (aarch64_sshr_n<mode>): Likewise.
11438         (aarch64_ushr_n<mode>): Likewise.
11439         (aarch64_<maxmin><mode>): Likewise.
11440         (aarch64_sqrt<mode>): Likewise.
11441         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
11442         (vshr<q>_n_*): Likewise.
11444 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11446         * config/aarch64/aarch64-builtins.c
11447         (aarch64_simd_builtin_type_mode): Handle SF types.
11448         (sf_UP): Define.
11449         (BUILTIN_GPF): Define.
11450         (aarch64_init_simd_builtins): Handle SF types.
11451         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
11452         (frecps): Likewise.
11453         (frecpx): Likewise.
11454         * config/aarch64/aarch64-simd.md
11455         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
11456         (aarch64_frecpe<mode>): New.
11457         (aarch64_frecps<mode>): Likewise.
11458         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
11459         (v8type): Add frecp<esx>.
11460         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
11461         (aarch64_frecps<mode>): Likewise.
11462         * config/aarch64/iterators.md (FRECP): New.
11463         (frecp_suffix): Likewise.
11464         * config/aarch64/arm_neon.h
11465         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
11467 2013-04-22  Christian Bruel  <christian.bruel@st.com>
11469         PR target/56995
11470         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
11471         (REG_CLASS_NAMES): Idem.
11472         (REG_CLASS_CONTENTS): Idem.
11473         (REGCLASS_HAS_FP_REG): Idem.
11474         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
11475         (sh_conditional_register_usage): Idem.
11477 2013-04-21  Jeff Law  <law@redhat.com>
11479         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
11480         (ssa_forward_propagate_and_combine): Use it.
11482 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11484         * lra.c: Update the flow chart diagram.
11486 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11488         PR rtl-optimization/56847
11489         * lra-constraints.c (process_alt_operands): Discourage alternative
11490         with non-matche doffsettable memory constraint fro memory with
11491         known offset.
11493 2013-04-19  Richard Biener  <rguenther@suse.de>
11495         PR tree-optimization/56982
11496         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
11497         function.
11498         * gimplify.c (gimplify_call_expr): Notice special calls.
11499         (gimplify_modify_expr): Likewise.
11500         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
11501         abnormal control flow receivers.
11502         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
11503         in the same way as cfun->has_nonlocal_labels.
11504         (gimple_purge_dead_abnormal_call_edges): Likewise.
11505         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
11506         receivers start a basic-block.
11508 2013-04-19  Richard Biener  <rguenther@suse.de>
11510         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
11511         member ...
11512         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
11513         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
11514         (SLP_TREE_LOAD_PERMUTATION): Add.
11515         (vect_transform_slp_perm_load): Adjust prototype.
11516         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
11517         (vect_free_slp_instance): Likewise.
11518         (vect_create_new_slp_node): Likewise.
11519         (vect_supported_slp_permutation_p): Remove.
11520         (vect_slp_rearrange_stmts): Adjust.
11521         (vect_supported_load_permutation_p): Likewise.  Inline
11522         vect_supported_slp_permutation_p here.
11523         (vect_analyze_slp_instance): Compute load permutations per
11524         slp node instead of per instance.
11525         (vect_get_slp_defs): Adjust.
11526         (vect_transform_slp_perm_load): Likewise.
11527         (vect_schedule_slp_instance): Remove redundant code.
11528         (vect_schedule_slp): Remove hack for PR56270, add it ...
11529         * tree-vect-stmts.c (vectorizable_load): ... here, do not
11530         CSE loads for SLP.  Adjust.
11532 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11534         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
11535         spelling in two comments.
11537 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11539         PR target/56797
11540         * config/arm/arm.c (load_multiple_sequence): Require SP
11541         as base register for loads if SP is in the register list.
11543 2013-04-19  Martin Jambor  <mjambor@suse.cz>
11545         PR tree-optimization/56718
11546         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11547         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11548         and made public.  Adjusted all callers.
11549         (ipa_intraprocedural_devirtualization): New function.
11550         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11551         (ipa_intraprocedural_devirtualization): Likewise.
11552         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11554 2013-04-19  Richard Biener  <rguenther@suse.de>
11556         PR tree-optimization/57000
11557         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11559 2013-04-19  Terry Guo  <terry.guo@arm.com>
11561         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11562         Replace with ...
11563         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
11564         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11565         (cortex_m4_fmacs): Use new reservations.
11566         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11568 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
11570         PR rtl-optimization/56999
11571         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11572         related code.
11573         (lra_coalesce): Remove split_origin_bitmap and related code.
11574         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
11575         ranges if necessary.
11577 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
11579         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11580         New array.
11581         (ix86_expand_call): Remove clobbered_registers array and use
11582         x86_64_ms_sysv_extra_clobbered_registers instead.
11583         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11584         Declare here.
11585         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11586         predicate.
11587         * config/i386/i386.md (*call_rex64_ms_sysv): Use
11588         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
11589         (*call_value_rex64_ms_sysv): Ditto.
11591 2013-04-18  Cary Coutant  <ccoutant@google.com>
11593         * dwarf2out.c (output_pubnames): Check die_perennial_p of
11594         parent instead of die_mark.
11596 2013-04-18  Diego Novillo  <dnovillo@google.com>
11598         * gimple.c (create_gimple_tmp): New.
11599         (get_expr_type): New.
11600         (build_assign): New.
11601         (build_type_cast): New.
11602         * gimple.h (enum ssa_mode): Define.
11603         (gimple_seq_set_location): New.
11604         * asan.c (build_check_stmt): Change some gimple_build_* calls
11605         to use build_assign and build_type_cast.
11607 2013-04-18  Richard Biener  <rguenther@suse.de>
11609         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11610         handle negative step.  Remove redundant checks.
11611         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11612         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11613         for negative step and grouped loads fail to vectorize.
11615 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11617         * emit-rtl.c (reset_insn_used_flags): New function.
11618         (reset_all_used_flags): Use it.
11619         (verify_insn_sharing): New function.
11620         (verify_rtl_sharing): Fix verification for SEQUENCEs.
11622 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
11624         PR tree-optimization/56984
11625         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11626         and (x >> M) >= N don't register any assertion if N << M is the
11627         minimum value.
11629 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11631         * lower-subreg.c (resolve_simple_move): If called self-recursive,
11632         do not delete_insn insns that have not yet been emitted, only
11633         unlink them with remove_insn.
11634         * df-scan.c (df_insn_delete): Revert r197492.
11636 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
11638         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11639         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11641 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11643         * config/arm/arm.md (movsicc_insn): Convert define_insn into
11644         define_insn_and_split.
11645         (and_scc,ior_scc,negscc): Likewise.
11646         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11648 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11650         * config/arm/arm.c (use_return_insn): Return 0 for targets that
11651         can benefit from using a sequence of LDRD instructions in epilogue
11652         instead of a single LDM instruction.
11654 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11656         PR 45688
11657         * doc/extend.texi: Fix typo.
11659 2013-04-17  Richard Biener  <rguenther@suse.de>
11661         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11662         (vect_build_slp_tree): ... here.
11663         (vect_build_slp_tree_1): Compute which stmts of the SLP group
11664         match.  Remove special-casing of mismatched complex loads.
11665         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11666         re-try the match with swapped commutative operands.
11667         (vect_supported_load_permutation_p): Remove special-casing of
11668         mismatched complex loads.
11669         (vect_analyze_slp_instance): Adjust.
11671 2013-04-17  Richard Biener  <rguenther@suse.de>
11673         PR rtl-optimization/56921
11674         * cfgloop.h (struct loop): Add simple_loop_desc member.
11675         (struct niter_desc): Mark with GTY(()).
11676         (simple_loop_desc): Do not use aux field but simple_loop_desc.
11677         * loop-iv.c (get_simple_loop_desc): Likewise.
11678         (free_simple_loop_desc): Likewise.
11680         Revert
11681         2013-04-16  Richard Biener  <rguenther@suse.de>
11683         PR rtl-optimization/56921
11684         * loop-init.c (pass_rtl_move_loop_invariants): Add
11685         TODO_do_not_ggc_collect to todo_flags_finish.
11686         (pass_rtl_unswitch): Same.
11687         (pass_rtl_unroll_and_peel_loops): Same.
11688         (pass_rtl_doloop): Same.
11690 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
11692         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11693         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11694         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11695         references.
11696         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11697         * tree-streamer.c (record_common_node): Adjust reference in comment.
11699 2013-04-17  Terry Guo  <terry.guo@arm.com>
11701         * config/arm/cortex-m4.md: Add a new bypass.
11703 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11705         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
11706         New pattern.
11707         (*subs_<optab><mode>_multp2): New pattern.
11708         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11709         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11711 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11713         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
11714         (*subs_mul_imm_<mode>): New pattern.
11716 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
11718         PR target/56948
11719         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
11720         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
11721         (vsx_movti_32bit): Change j->wa to O->wa.
11723 2013-04-16  Richard Biener  <rguenther@suse.de>
11725         PR rtl-optimization/56921
11726         * loop-init.c (pass_rtl_move_loop_invariants): Add
11727         TODO_do_not_ggc_collect to todo_flags_finish.
11728         (pass_rtl_unswitch): Same.
11729         (pass_rtl_unroll_and_peel_loops): Same.
11730         (pass_rtl_doloop): Same.
11732 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11734         * config/arm/arm.c (emit_multi_reg_push): New declaration
11735         for an existing function.
11736         (arm_emit_strd_push): New function.
11737         (arm_expand_prologue): Used here.
11738         (arm_emit_ldrd_pop): New function.
11739         (arm_expand_epilogue): Used here.
11740         (arm_get_frame_offsets): Update condition.
11741         (arm_emit_multi_reg_pop): Add a special case for load of a single
11742         register with writeback.
11744 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
11746         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
11747         description.
11749 2013-04-16  Richard Biener  <rguenther@suse.de>
11751         PR tree-optimization/56756
11752         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
11753         (first_mem_ref_loc): New.
11754         (execute_sm): Place the load temporarily before a previous
11755         access instead of in the latch edge to ensure its SSA dependencies
11756         are defined at points dominating the load.
11758 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
11760         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
11761         correct fix by moving header and footer insn to the footer of
11762         the merged basic block.  Clear BB_END of the merged-away block.
11764         PR middle-end/43631
11765         * emit-rtl.c (make_note_raw): New function.
11766         (link_insn_into_chain): New static inline function.
11767         (add_insn): Use it.
11768         (add_insn_before, add_insn_after): Factor insn chain linking code...
11769         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
11770         using link_insn_into_chain.
11771         (note_outside_basic_block_p): New helper function for emit_note_after
11772         and emit_note_before.
11773         (emit_note_after): Use nobb variant of add_insn_after if the note
11774         should not be contained in a basic block.
11775         (emit_note_before): Use nobb variant of add_insn_before if the note
11776         should not be contained in a basic block.
11777         (emit_note_copy): Use make_note_raw.
11778         (emit_note): Likewise.
11779         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
11780         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
11781         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
11782         the moved barrier the tail of the basic block it follows.
11783         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
11785 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
11787         PR tree-optimization/56962
11788         * gimple-ssa-strength-reduction.c (record_increment): Only set
11789         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
11790         either rhs1 or rhs2 is equal to c->base_expr.
11792 2013-04-15  Richard Biener  <rguenther@suse.de>
11794         PR tree-optimization/56933
11795         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
11796         member.
11797         (GROUP_READ_WRITE_DEPENDENCE): Remove.
11798         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
11799         * tree-vect-data-refs.c (vect_analyze_group_access): Move
11800         dependence check ...
11801         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11802         ... here.
11803         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11804         GROUP_READ_WRITE_DEPENDENCE.
11806 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11808         * emit-rtl.c (reset_all_used_flags): New function.
11809         (verify_rtl_sharing): Call reset_all_used_flags before and after
11810         performing the checks.
11812 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11814         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11815         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11816         * config/arm/constraints.md (De): New constraint.
11817         * config/arm/neon.md (anddi3_neon): Delete.
11818         (neon_vand<mode>): Expand to standard anddi3 pattern.
11819         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11820         Move earlier in the file.
11821         (neon_inv_logic_op2): Likewise.
11822         (arm_anddi_operand_neon): New predicate.
11824 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11826         * configure.ac (gcc_cv_ld_as_needed): Set
11827         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11828         Use -z ignore, -z record on *-*-solaris2*.
11829         (HAVE_LD_AS_NEEDED): Update comment.
11830         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11831         * configure: Regenerate.
11832         * config.in: Regenerate.
11833         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11834         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11835         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11836         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11837         equivalents.  Fix markup.
11838         * doc/tm.texi: Regenerate.
11840 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
11842         * config/i386/i386.opt: New option mstack-protector-guard=.
11843         * config/i386/i386-opts.h: Add enum stack_protector_guard.
11844         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11845         TARGET_SSP_TLS_GUARD.
11846         * config/i386/i386.c (ix86_option_override_internal): Set
11847         ix86_stack_protector_guard.
11848         * config/i386/i386.md (stack_protect_set): Enable for
11849         TARGET_SSP_TLS_GUARD only.
11850         (stack_protect_set_<mode>): Ditto.
11851         (stack_protect_test): Ditto.
11852         (stack_protect_test_<mode>): Ditto.
11853         * doc/invoke.texi (i386 Option): Document.
11855 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
11857         PR target/56890
11858         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11859         (S_MODES): Set H_MODE bit.
11860         (SF_MODES): Set only S_MODE and SF_MODE bits.
11861         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11862         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11863         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11864         <MODE_FLOAT>: Likewise.
11866 2013-04-15  Joey Ye  <joey.ye@arm.com>
11868         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11870 2013-04-15  Joey Ye  <joey.ye@arm.com>
11872         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11873         for real far jump.
11874         (thumb_far_jump_used_p): Count instruction size and set
11875         far_jump_used.
11877 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
11879         * reorg.c (fill_simple_delay_slots): Reindent block of code.
11880         * resource.c (mark_target_live_regs): Reformat conditional block.
11882 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11884         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11885         notes, they are emitted only just before final.
11886         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11888 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11890         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11891         * cfgrtl.c (delete_insn): Call it here instead.
11892         * lra-spills.c (lra_final_code_change): Use delete_insn.
11893         * haifa-sched.c (sched_remove_insn): Likewise.
11894         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11895         returning to the nop pool.
11896         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11897         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
11899 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11901         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11902         * doc/tm.texi: Regenerated.
11904 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
11906         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11907         QImode checks.
11909 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11911         * df-core.c (df_find_def): Compare register numbers.
11912         (df_find_use): Likewise.
11914 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
11916         PR target/56903
11917         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11918         lra_in_progress for return.
11920 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11922         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11923         define_insn into define_insn_and_split and emit movsicc patterns.
11925 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11927         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11929 2013-04-12  Richard Biener  <rguenther@suse.de>
11931         * tree-pass.h (TODO_do_not_ggc_collect): New.
11932         * passes.c (execute_one_ipa_transform_pass): Honor
11933         TODO_do_not_ggc_collect.
11934         (execute_one_pass): Likewise.
11936         Revert
11937         2013-04-10  Richard Biener  <rguenther@suse.de>
11939         * passes.c (init_optimization_passes): Remove reload pass.
11940         * ira.c (do_reload): Merge into ...
11941         (ira): ... this.
11942         (rest_of_handle_reload): Remove.
11943         (pass_reload): Likewise.
11944         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11945         of reload for vzeroupper pass placement.
11947 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
11949         PR tree-optimization/56918
11950         PR tree-optimization/56920
11951         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11952         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11953         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11954         use wide_mul_with_sign method.
11956 2013-04-12  Richard Biener  <rguenther@suse.de>
11958         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11959         not be considered a gimple constant.
11961 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11963         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11964         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11965         operations.
11967 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11968             Jakub Jelinek  <jakub@redhat.com>
11970         * opts.c: Include diagnostic-color.h.
11971         (common_handle_option): Handle OPT_fdiagnostics_color_.
11972         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11973         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11974         (diagnostic-color.o): New.
11975         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11976         (diagnostic_color_rule): New enum.
11977         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11978         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11979         the location string.
11980         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11981         either NULL, or color kind.
11982         * diagnostic-color.c: New file.
11983         * diagnostic-color.h: New file.
11984         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11985         arguments.
11986         * doc/invoke.texi (-fdiagnostics-color): Document.
11987         * pretty-print.h (pp_show_color): Define.
11988         (struct pretty_print_info): Add show_color field.
11989         * diagnostic.c: Include diagnostic-color.h.
11990         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11991         macros.  Colorize error:, warning: etc. strings and also the location
11992         string.
11993         (diagnostic_show_locus): Colorize the caret line.
11994         * pretty-print.c: Include diagnostic-color.h.
11995         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
11996         inside of %< %> quotes or quoted through q format modifier.
11998 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12000         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12002 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12004         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12005         code in CC_NZ mode.
12006         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12007         pattern.
12009 2013-04-11  Marek Polacek  <polacek@redhat.com>
12011         PR tree-optimization/48184
12012         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12014 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
12016         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12017         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12018         (skip_simple_arithmetic): Tidy up.
12019         * tree.h (skip_simple_constant_arithmetic): Declare.
12021 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12023         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12025 2013-04-11  Richard Biener  <rguenther@suse.de>
12027         * tree-vect-loop.c (get_initial_def_for_induction): Properly
12028         generate vector constants.
12030 2013-04-11  Richard Biener  <rguenther@suse.de>
12032         PR tree-optimization/56878
12033         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12034         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12035         New function.
12036         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12037         Prefer to align the DR with the most invariant base address.
12039 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12041         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12042         comment.
12044 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
12046         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12047         floating-point vector comparisons against 0.
12049 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
12051         PR tree-optimization/56899
12052         * fold-const.c (extract_muldiv_1): Apply distributive law
12053         only if TYPE_OVERFLOW_WRAPS (ctype).
12055 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
12057         PR target/56124
12058         * ira-costs.c (scan_one_insn): Check whether the source rtx of
12059         loading has side effect.
12061 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12063         * config/sparc/sparc.c: Include tree-pass.h.
12064         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12065         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12066         head of file.  Change return type.  Split off gate function.
12067         (sparc_gate_work_around_errata): New function.
12068         (pass_work_around_errata): New pass definition.
12069         (insert_pass_work_around_errata) New pass insert definition to
12070         insert pass_work_around_errata just after delayed-branch scheduling.
12071         (sparc_option_override): Insert the pass.
12072         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12074 2013-04-10  David S. Miller  <davem@davemloft.net>
12076         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12077         or -mcpu=hypersparc.
12079         * target.def (cstore_mode): New hook.
12080         * target.h: Include insn-codes.h
12081         * targhooks.c: Likewise.
12082         (default_cstore_mode): New function.
12083         * targhooks.h: Declare it.
12084         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12085         * doc/tm.texi: Rebuild.
12086         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12087         target hook, rather than inspecting the insn_data.
12088         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12089         (TARGET_CSTORE_MODE): Redefine.
12090         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12091         result patterns.
12092         * config/sparc/predicates.md (cstore_result_operand): New special
12093         predicate.
12094         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12095         Use it for operand 0.
12096         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12097         (*snesi_special): Likewise.
12098         (*snesi_zero): Likewise.
12099         (*seqsi_zero): Likewise.
12100         (*sltu_insn): Likewise.
12101         (*sgeu_insn): Likewise.
12102         (*seqdi_special): Make operand 0 and comparison operation be of
12103         DImode.
12104         (*snedi_special): Likewise.
12105         (*snedi_special_vis3): Likewise.
12106         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12107         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12108         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12109         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12110         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12111         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12112         (*sltu_extend_sp64): Likewise.
12113         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12114         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12115         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12116         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12117         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12119 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
12121         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12122         (aarch64_start_file): Use the new function.
12124 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12125             Jason Merrill  <jason@redhat.com>
12127         * common.opt: Add -gdwarf.
12128         * opts.c (common_handle_option): Handle it.
12129         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12131 2013-04-10  Richard Biener  <rguenther@suse.de>
12133         * passes.c (execute_todo): Do not call ggc_collect conditional here.
12134         (execute_one_ipa_transform_pass): But unconditionally here.
12135         (execute_one_pass): And here.
12136         (init_optimization_passes): Remove reload pass.
12137         * tree-pass.h (TODO_ggc_collect): Remove.
12138         (pass_reload): Likewise.
12139         * ira.c (do_reload): Merge into ...
12140         (ira): ... this.
12141         (rest_of_handle_reload): Remove.
12142         (pass_reload): Likewise.
12143         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12144         of reload for vzeroupper pass placement.
12145         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12146         and todo_flags_finish of all passes.
12148 2013-04-10  Richard Biener  <rguenther@suse.de>
12150         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12151         first_const_oprnd field, rename first_def_type to first_op_type.
12152         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12153         (vect_get_and_check_slp_defs): Always use the type of the
12154         operand.  Allow mixed vect_external_def, vect_constant_def types.
12155         (vect_get_constant_vectors): Handle mixed vect_external_def,
12156         vect_constant_def types.
12158 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
12160         PR tree-optimization/55524
12161         * tree-ssa-math-opts.c
12162         (convert_mult_to_fma): Don't use an fms construct
12163         when we don't have an fms operation, but fnma, and it looks
12164         likely that we'll be able to use the latter.
12166 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
12168         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12169         function.
12170         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12171         inline fail caused by overwritable functions.
12173 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
12175         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12176         unnecessary bits in the constant power of two case.
12178 2013-04-10  Richard Biener  <rguenther@suse.de>
12180         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12181         broken code swapping operands.
12182         (vect_build_slp_tree): Do not compute load permutations here.
12183         (vect_analyze_slp_instance): Compute load permutations here,
12184         after building the SLP tree.
12186 2013-04-09  Christian Bruel  <christian.bruel@st.com>
12188         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12189         of next/prev_real_insn.
12191 2013-04-09  Jan Hubicka  <jh@suse.cz>
12193         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12194         Drop aliased parameter.
12195         (function_and_variable_visibility): Do not handle alias pairs.
12196         * cgraph.c (varpool_externally_visible_p): Update prototype.
12197         * varpool.c (varpool_add_new_variable): Update.
12199 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12201         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12203 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
12205         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12207         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12209 2013-04-09  Marek Polacek  <polacek@redhat.com>
12211         PR tree-optimization/48762
12212         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12214 2013-04-09  Richard Biener  <rguenther@suse.de>
12216         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12217         dealing with cost.
12218         (vect_build_slp_tree): Likewise.
12219         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12220         calculating the cost of a SLP instance.
12221         (vect_analyze_slp_instance): Use it from here, after building
12222         the SLP tree.
12224 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12226         PR middle-end/56883
12227         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12228         expand_omp_for_static_chunk): Use simple_p = true in
12229         force_gimple_operand_gsi calls when assigning to addressable decls.
12231 2013-04-09  Jeff Law  <law@redhat.com>
12233         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12234         when the boolean was created by converting a wider object which
12235         had a boolean range.
12237 2013-04-09  Richard Biener  <rguenther@suse.de>
12239         * tree-vectorizer.h (slp_void_p): Remove.
12240         (slp_tree): Typedef before _slp_tree declaration.
12241         (struct _slp_tree): Use a vector of slp_tree as children.
12242         (vect_get_place_in_interleaving_chain): Remove.
12243         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12244         Move ...
12245         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12246         and make static.
12247         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12248         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12249         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12250         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12251         Use slp_node instead of slp_void_p and adjust.
12253 2013-04-09  Richard Biener  <rguenther@suse.de>
12255         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12256         work that is not necessary.
12258 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12260         PR tree-optimization/56854
12261         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12262         forward into clobber stmts if it would change MEM_REF lhs into
12263         non-MEM_REF.
12265 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
12267         * tree.c (type_hash_lookup, type_hash_add): Make static.
12268         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12270 2013-04-09  Richard Biener  <rguenther@suse.de>
12272         * tree.h (unsave_expr_now): Remove.
12273         * tree-inline.c (mark_local_for_remap_r): Remove.
12274         (unsave_expr_1): Likewise.
12275         (unsave_r): Likewise.
12276         (unsave_expr_now): Likewise.
12277         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12278         (propagate_tree_value): Likewise.
12280 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
12282         * doc/rtl.texi (sequence): Rewrite documentation to match the
12283         current use of SEQUENCE rtl objects.
12284         * rtl.def (SEQUENCE): Likewise.
12286         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12287         Update documentation.
12288         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12289         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12291         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12293 2013-04-08  Teresa Johnson  <tejohnson@google.com>
12295         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12296         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12297         methods.
12298         (estimate_edge_size_and_time): Add comment to suggest using rounding
12299         methods.
12300         (estimate_node_size_and_time): Ditto.
12301         (remap_edge_change_prob): Use helper rounding divide methods.
12302         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
12303         (gimple_mod_pow2_value_transform): Ditto.
12304         (gimple_mod_subtract_transform): Ditto.
12305         (gimple_ic_transform): Ditto.
12306         (gimple_stringops_transform): Ditto.
12307         * stmt.c (conditional_probability): Ditto.
12308         (emit_case_dispatch_table): Ditto.
12309         * lto-cgraph.c (merge_profile_summaries): Ditto.
12310         * tree-optimize.c (execute_fixup_cfg): Ditto.
12311         * cfgcleanup.c (try_forward_edges): Ditto.
12312         * cfgloopmanip.c (scale_loop_profile): Ditto.
12313         (loopify): Ditto.
12314         (duplicate_loop_to_header_edge): Ditto.
12315         (lv_adjust_loop_entry_edge): Ditto.
12316         * tree-vect-loop.c (vect_transform_loop): Ditto.
12317         * profile.c (compute_branch_probabilities): Ditto.
12318         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
12319         * lto-streamer-in.c (input_cfg): Ditto.
12320         * gimple-streamer-in.c (input_bb): Ditto.
12321         * ipa-cp.c (update_profiling_info): Ditto.
12322         (update_specialized_profile): Ditto.
12323         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
12324         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
12325         rounding methods.
12326         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12327         (compute_trg_info): Ditto.
12328         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
12329         (purge_dead_edges): Ditto.
12330         * loop-unswitch.c (unswitch_loop): Ditto.
12331         * cgraphclones.c (cgraph_clone_edge): Ditto.
12332         (cgraph_clone_node): Ditto.
12333         * tree-inline.c (copy_bb): Ditto.
12334         (copy_edges_for_bb): Ditto.
12335         (initialize_cfun): Ditto.
12336         (copy_cfg_body): Ditto.
12337         (expand_call_inline): Ditto.
12339 2013-04-08  Kai Tietz  <ktietz@redhat.com>
12341         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
12342         TARGET_CYGWIN64 by TARGET_64BIT.
12344 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12346         * config/epiphany/epiphany.md (GPR_1): New constant.
12347         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
12348         * config/epiphany/epiphany.c (gen_compare_reg):
12349         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
12350         is already in place.
12351         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
12352         Don't require being called during rtl expansion; If y operlaps r0,
12353         return 0.
12354         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
12355         (epiphany_expand_epilogue): Likewise.
12357         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
12358         Don't use CC_FPmode for ORDERED / UNORDERED.
12359         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
12361         * config/epiphany/constraints.md (CnL): New constraint.
12362         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
12363         * config/epiphany/predicates.md (add_operand): Allow 1024.
12365         * config/epiphany/epiphany.md (logical_op): New code iterator.
12366         (op_mnc): New code attribute.
12367         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
12368         (mov_f+1, mov_f+2): New peephole2 patterns.
12370         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
12371         (cstoresi4): Also allow re-use of zero result when doing a NE
12372         comparison to a non-zero operand.
12373         Use (clobber (scratch)) for first insn if the gpr output is not needed.
12375         * config/epiphany/epiphany.md (<insn_opname>v2si3):
12376         Use gen_addsi3_i / gen_subsi3_i.
12378 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12380         PR c++/34949
12381         PR c++/50243
12382         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
12383         contain anything but clobbers, at most one __builtin_stack_restore,
12384         optionally debug stmts and final resx, and if it has at least one
12385         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
12386         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
12387         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
12388         which isn't defaut definition, remove them.
12389         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
12390         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
12391         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
12392         with MEM_REF LHS with SSA_NAME address.
12394 2013-04-08  Jeff Law  <law@redhat.com>
12396         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
12398 2013-04-08  Richard Biener  <rguenther@suse.de>
12400         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
12401         extra newline.
12402         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
12403         determined vector type.
12404         (vect_analyze_data_refs): Likewise.
12405         (vect_get_new_vect_var): Adjust.
12406         (vect_create_destination_var): Preserve SSA name versions.
12407         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
12408         not dump anything here.
12410 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12412         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
12413         Add member lr_slot_known.
12414         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
12415         if necessary.
12416         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12417         Remove code that sets lr_slot_offset according to what a previous
12418         version of epiphany_emit_save_restore used to do.
12419         (epiphany_emit_save_restore): When doing an lr save or restore,
12420         set/verify lr_slot_known and lr_slot_offset.
12422 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
12424         PR target/54338
12425         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
12426         in ALL_REGS.
12428 2013-04-08  Richard Biener  <rguenther@suse.de>
12430         * alias.c (find_base_term): Fix thinko in previous change.
12432 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12434         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
12435         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
12436         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
12437         if possible to compute val.
12438         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
12439         For QImode integers don't require anything about precision.  Use
12440         const_with_all_bytes_same to find out if the constant doesn't have
12441         repeated bytes in it.
12443 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12445         * config/s390/s390.c (s390_expand_insv): Only accept insertions
12446         within mode size.
12448 2013-04-08  Marek Polacek  <polacek@redhat.com>
12450         PR rtl-optimization/48182
12451         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
12452         value to 1.
12454 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12456         PR target/55487
12457         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
12458         nuses, make sure we have a label.
12460 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12462         PR target/56843
12463         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
12464         (rs6000_emit_swdiv_low_precision): Remove.
12465         (rs6000_emit_swdiv): Rewrite to handle between one and four
12466         iterations of Newton-Raphson generally; modify required number of
12467         iterations for some cases.
12468         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
12470 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
12472         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
12473         set-but-unused variable.
12475         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
12476         basic blocks of released function bodies garbage-collectable.
12478         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
12479         (struct rtl_opt_pass): Add TODO_df_finish.
12481         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
12483 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12485         * config/arm/constraints.md (q): New constraint.
12486         * config/arm/ldrdstrd.md: New file.
12487         * config/arm/arm.md (ldrdstrd.md) New include.
12488         (arm_movdi): Use "q" instead of "r" constraint
12489         for double-word memory access.
12490         (movdf_soft_insn): Likewise.
12491         * config/arm/vfp.md (movdi_vfp): Likewise.
12492         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
12493         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
12494         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
12495         (mem_ok_for_ldrd_strd): Likewise.
12496         (output_move_double): Update assertion.
12498 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12500         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
12502 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12504         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
12505         define_insn_and_split.
12506         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
12508 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12510         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
12511         define_insn_and_split.
12512         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
12513         (shiftsi3_compare): New pattern.
12514         (rrx): New pattern.
12515         * config/arm/unspecs.md (UNSPEC_RRX): New.
12517 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12519         * config/arm/arm.md (negdi_extendsidi): New pattern.
12520         (negdi_zero_extendsidi): Likewise.
12522 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12524         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
12525         define_insn_and_split.
12526         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
12527         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
12529 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12531         * config/arm/arm.md (arm_subdi3): Convert define_insn into
12532         define_insn_and_split.
12533         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
12534         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
12536 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12538         * config/arm/arm.md (subsi3_carryin): New pattern.
12539         (subsi3_carryin_const): Likewise.
12540         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
12541         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12543 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12545         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12547 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12549         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12550         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12552 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12554         * config/arm/arm.c (arm_expand_builtin): Change fcode
12555         type to unsigned int.
12557 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12559         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12561 2013-04-04  Ian Lance Taylor  <iant@google.com>
12563         * doc/standards.texi (Standards): The Go frontend supports the Go 1
12564         language standard.
12566 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
12568         PR middle-end/56729
12569         * df-scan.c (df_insn_delete): Disable failing assert.
12571 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12573         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12574         New function prototype.
12575         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12576         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12577         (arm_builtin_vectorized_function): New function.
12579 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12581         * config/arm/arm_neon_builtins.def: New file.
12582         * config/arm/arm.c (neon_builtin_data): Move contents to
12583         arm_neon_builtins.def.
12584         (enum arm_builtins): Include neon builtin definitions.
12585         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12586         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12588 2013-04-04  Marek Polacek  <polacek@redhat.com>
12590         PR tree-optimization/48186
12591         * predict.c (maybe_hot_frequency_p): Return false if
12592         HOT_BB_FREQUENCY_FRACTION is 0.
12593         (cgraph_maybe_hot_edge_p): Likewise.
12595 2013-04-04  Richard Biener  <rguenther@suse.de>
12597         PR tree-optimization/56826
12598         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12599         more accurately.
12601 2013-04-04  Richard Biener  <rguenther@suse.de>
12603         PR tree-optimization/56213
12604         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12605         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12607 2013-04-04  Richard Biener  <rguenther@suse.de>
12609         PR tree-optimization/56837
12610         * tree-loop-distribution.c (classify_partition): For non-zero
12611         values require that the value has the same precision as its
12612         mode to be useful as memset value.
12614 2013-04-03  Nick Clifton  <nickc@redhat.com>
12616         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12617         (fmssf4): Use fmsf.s on E3V5 architectures.
12618         (fnmasf4): Use fnmaf.s on E3V5 architectures.
12619         (fnmssf4): Use fnmsf.s on E3V5 architectures.
12621 2013-04-03  Jeff Law  <law@redhat.com>
12623         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12624         (lra-eliminations.o): Likewise.
12626 2013-04-03  Teresa Johnson  <tejohnson@google.com>
12628         * gcov-io.c (compute_working_sets): Moved most of body of old
12629         compute_working_sets here from profile.c.
12630         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12631         (gcov_working_set_t): Moved typedef here from basic-block.h
12632         (compute_working_set): Declare.
12633         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12634         (get_working_sets): Renamed from compute_working_set,
12635         replace most of body with call to new compute_working_sets.
12636         (get_exec_counts): Replace call to compute_working_sets
12637         to get_working_sets.
12638         * profile.h (get_working_sets): Renamed from compute_working_set.
12639         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12640         to get_working_sets.
12641         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12642         * gcov-dump.c (dump_working_sets): New function.
12644 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
12646         * hwint.c (sext_hwi, zext_hwi): New functions.
12647         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12648         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12649         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12650         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12651         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12652         (sext_hwi, zext_hwi): New functions.
12654 2013-04-03  Jeff Law  <law@redhat.com>
12656         PR tree-optimization/56799
12657         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12658         back test for widening conversion erroneously dropped in prior change.
12660 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12662         PR target/56809
12663         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12664         instead of next_real_insn.
12666 2013-04-03  Marek Polacek  <polacek@redhat.com>
12668         PR sanitizer/55702
12669         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12671 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12673         PR target/56809
12674         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12675         next_real_insn.
12676         (thumb1_output_casesi): Likewise.
12677         (thumb2_output_casesi): Likewise.
12679 2013-04-03  Richard Biener  <rguenther@suse.de>
12681         PR tree-optimization/56817
12682         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12683         Split out ...
12684         (tree_unroll_loops_completely_1): ... new function to manually
12685         walk the loop tree, properly defering outer loops of unrolled
12686         loops to later iterations.
12688 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12690         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12691         (vectorizable_load): Likewise.
12692         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12693         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12695 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12697         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12698         BIT_FIELD_REF.
12700 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12702         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12704 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
12706         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
12708 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12710         PR tree-optimization/56790
12711         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
12712         folding.
12714 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12716         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
12717         Handle VEC_MERGE.
12718         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
12719         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
12720         equal arguments.
12722 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
12724         PR c/19449
12725         * tree.h (force_folding_builtin_constant_p): New decl.
12726         * builtins.c (force_folding_builtin_constant_p): New variable.
12727         (fold_builtin_constant_p): Fold immediately also if
12728         force_folding_builtin_constant_p.
12730 2013-04-03  Richard Biener  <rguenther@suse.de>
12732         PR tree-optimization/56812
12733         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
12734         DRs of the same interleaving chain are independent.
12736 2013-04-02  Jason Merrill  <jason@redhat.com>
12738         * gdbinit.in (pbb): Use debug fn.
12740 2013-04-02  Lawrence Crowl  <crowl@google.com>
12742         * sese.h (struct ivtype_map_elt_s): Remove unused.
12743         (extern debug_ivtype_map): Remove unused.
12744         (extern eq_ivtype_map_elts): Remove unused.
12745         * sese.c (debug_ivtype_map): Removed unused.
12746         (debug_ivtype_map_1): Removed unused.
12747         (debug_ivtype_elt): Remove unused.
12748         (eq_ivtype_map_elts): Remove unused.
12751 2013-04-02  Kai Tietz  <ktietz@redhat.com>
12753         PR target/52790
12754         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
12755         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
12756         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
12757         function.
12758         (legitimize_pe_coff_symbol): Likewise.
12759         (is_imported_p): New helper-function.
12760         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
12761         for Windows x64 targets.
12762         (ix86_expand_prologue): Optimize for pe-coff targets.
12763         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
12764         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
12765         medium/large code-model.
12766         (legitimize_pic_address): Likewise.
12767         (legitimize_tls_address): Likewise.
12768         (ix86_expand_call): Likewise.
12769         (x86_output_mi_thunk): Likewise.
12770         (get_dllimport_decl): Add new beimport argument.
12771         (construct_plt_address): Don't assert for x64 pe-coff targets.
12772         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
12773         targets.
12774         (SYMBOL_FLAG_STUBVAR): New macro.
12775         (SYMBOL_REF_STUBVAR_P): Likewise.
12776         * config/i386/winnt.c (stub_list): New structure.
12777         (stub_head): New local variable.
12778         (i386_pe_record_stub): New function.
12779         (i386_pe_file_end): Emit refptr-stubs.
12781 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
12783         PR rtl-optimization/56745
12784         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
12785         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
12787         PR c++/34949
12788         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
12789         and both of them are MEM_REFs, just compare first argument for
12790         equality and attempt to deal even with differing offsets.
12792         PR c++/34949
12793         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
12794         of gimple_clobber_p to be MEM_REF.
12795         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
12796         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
12797         after gimplification.
12798         * asan.c (get_mem_ref_of_assignment): Don't instrument
12799         gimple_clobber_p stmts.
12800         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
12801         gimple_clobber_p stmt if they have MEM_REF lhs and
12802         are dead because of another gimple_clobber_p stmt.
12803         * tree-ssa-live.c (clear_unused_block_pointer): Treat
12804         gimple_clobber_p stmts like debug stmts.
12805         (remove_unused_locals): Remove clobbers with MEM_REF lhs
12806         that refer to unused VAR_DECLs or uninitialized values.
12807         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12808         gimple_clobber_p stmts if they refer to removed parameters.
12809         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12810         formatting.
12812 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
12814         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12815         using SWI48 mode attribute.
12817 2013-04-02  Wei Mi  <wmi@google.com>
12819         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12820         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12821         *<rotate_insn><mode>3_mask in i386.md.
12823 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12825         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12827 2013-04-02  Richard Biener  <rguenther@suse.de>
12829         PR tree-optimization/56778
12830         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12831         Runtime alias tests are not supported for gather loads.
12832         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12833         stmts referenced from SSA operands before updating SSA form.
12835 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
12836             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12838         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12839         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12840         * config/arm/cortex-a53.md: New file.
12841         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12842         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12843         * config/arm/arm.c (arm_issue_rate): Likewise.
12844         * config/arm/arm-tune.md: Regenerate
12845         * config/arm/arm-tables.opt: Regenerate.
12846         * config/arm/arm-cores.def: Add cortex-a53.
12848 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
12850         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12851         non-static link.
12853 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12855         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12856         scalar load/store operations using B/H registers.
12857         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12859 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12861         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12862         scalar move.
12863         * config/aarch64/aarch64.c
12864         (aarch64_simd_scalar_immediate_valid_for_move): New.
12865         * config/aarch64/aarch64-protos.h
12866         (aarch64_simd_scalar_immediate_valid_for_move): New.
12867         * config/aarch64/constraints.md (Dh, Dq): New.
12868         * config/aarch64/iterators.md (hq): New.
12870 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
12872         * reorg.c (get_branch_condition): Deal with conditional returns.
12873         (fill_simple_delay_slots): Remove dead code dealing with jumps.
12875 2013-04-01  Wei Mi  <wmi@google.com>
12877         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12878         Truncate operand 2 using %b asm operand modifier.
12879         (*<shift_insn><mode>3_mask): Ditto.
12880         (*<rotate_insn><mode>3_mask): Ditto.
12882 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
12884         PR middle-end/56798
12885         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12887 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
12889         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12890         of next_real_insn.
12891         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12893 2013-03-30  Lawrence Crowl  <crowl@google.com>
12895         * dse.c (clear_alias_sets): Remove never set.
12896         (disqualified_clear_alias_sets): Remove never set.
12897         (clear_alias_mode_pool): Remove never set.
12898         (dse_step0): Remove condition that is never true.
12899         (canon_address): Remove condition that is never true.
12900         (dse_step7): Remove condition that is never true.
12901         (rest_of_handle_dse): Remove condition that is never true.
12902         (rest_of_handle_dse::did_global): Remove never read from above.
12903         (dse_step2_spill): Remove never called from above.
12904         (dse_step5_spill): Remove never called from above.
12906 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
12908         * doc/md.texi (Standard Names) <casesi>: Update documentation for
12909         JUMP_TABLE_DATA changes.
12910         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12911         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12912         (Insns) <jump_table_data>: New entry.
12913         * doc/tm.texi: Regenerate.
12915         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12917         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12918         for table jump at the end of a basic block using tablejump_p.
12919         * targhooks.c (default_invalid_within_doloop): Likewise.
12920         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12921         target hook implementation that is identical to the default hook.
12922         (rs6000_invalid_within_doloop): Remove.
12924         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12925         unused variable from tablejump_p call.
12927         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12928         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12929         (INSN_DELETED_P): Likewise.
12930         (emit_jump_table_data): New prototype.
12931         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12932         after 4th as unused.
12933         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12934         * sched-vis.c (print_insn): Likewise.
12935         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12936         insn for compatibility with back ends that use next_active_insn to
12937         identify jump table data.
12938         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12939         (remove_insn): Likewise.
12940         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12941         to be emitted.
12942         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12943         (emit_jump_table_data): New function.
12945         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12946         basic block, a JUMP_TABLE_DATA never is.
12947         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12948         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12949         off from code handling real insns.
12950         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12951         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12952         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12953         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12954         is not a NONDEBUG_INSN_P.
12955         * ira-costs.c (scan_one_insn): Likewise.
12956         * jump.c (mark_all_labels): Likewise.
12957         (mark_jump_label_1): Likewise.
12958         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12959         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12960         a basic block.
12961         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12962         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12963         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12964         JUMP_TABLE_DATA_P insns.
12965         (calculate_elim_costs_all_insns): Likewise.
12966         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12967         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12968         (delete_output_reload): Code style fixups.
12969         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12970         insn flags on this non-insn.
12971         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12972         as scheduling barriers, for pre-change compatibility.
12973         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12974         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12976         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12977         redundant JUMP_TABLE_DATA_P test.
12978         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12979         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12980         (frv_for_each_packet): Likewise.
12981         * config/i386/i386.c (min_insn_size): Likewise.
12982         (ix86_avoid_jump_mispredicts): Likewise.
12983         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12984         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12985         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12986         (mips16_insn_length): Robustify.
12987         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12988         (mips16_split_long_branches): Likewise.
12989         * config/pa/pa.c (pa_combine_instructions): Likewise.
12990         * config/rs6000/rs6000.c (get_next_active_insn): Treat
12991         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12992         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12993         as contributing to pool range lengths.
12994         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12995         Remove redundant JUMP_TABLE_DATA_P test.
12996         (sh_loop_align): Likewise.
12997         (split_branches): Likewise.
12998         (sh_insn_length_adjustment): Likewise.
12999         * config/spu/spu.c (get_branch_target): Likewise.
13001 2013-03-29  Jan Hubicka  <jh@suse.cz>
13003         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13004         gcov streaming; stream hot bb threshold to ltrans.
13005         * predict.c (get_hot_bb_threshold): Break out from ....
13006         (maybe_hot_count_p): ... here.
13007         (set_hot_bb_threshold): New function.
13008         * lto-section-in.c (lto_section_name): Add profile.
13009         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13010         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13011         and data-streamer.h
13012         (histogram_entry): New structure.
13013         (histogram, histogram_pool): New global vars.
13014         (histogram_hash): New structure.
13015         (histogram_hash::hash): New method.
13016         (histogram_hash::equal): Likewise.
13017         (account_time_size): New function.
13018         (cmp_counts): New function.
13019         (dump_histogram): New function.
13020         (ipa_profile_generate_summary): New function.
13021         (ipa_profile_write_summary): New function.
13022         (ipa_profile_read_summary): New function.
13023         (ipa_profile): Decide on threshold.
13024         (pass_ipa_profile): Add ipa_profile_write_summary and
13025         ipa_profile_read_summary.
13026         * Makefile.in (ipa.o): Update dependencies.
13027         * lto-streamer.h (LTO_section_ipa_profile): New section.
13029 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13031         * tree.h (VAR_P): New.
13033 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
13035         PR lto/56777
13036         * doc/invoke.texi ([-fwhole-program]): Fix typo.
13038 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
13040         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13041         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13042         (control_flow_insn_p): Likewise.
13043         * cfgrtl.c (duplicate_insn_chain): Likewise.
13044         * final.c (get_attr_length_1): Likewise.
13045         (shorten_branches): Likewise.
13046         (final_scan_insn): Likewise.
13047         * function.c (instantiate_virtual_regs): Likewise.
13048         * gcse.c (insert_insn_end_basic_block): Likewise.
13049         * ira-costs.c (scan_one_insn): Likewise.
13050         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13051         * lra.c (check_rtl): Likewise.
13052         * reload1.c (elimination_costs_in_insn): Likewise.
13053         * reorg.c (follow_jumps): Likewise.
13055         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13056         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13057         (thumb_far_jump_used_p): Likewise.
13058         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13059         (workaround_speculation): Likewise.
13060         (add_sched_insns_for_speculation): Likewise.
13061         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13062         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13063         (frv_for_each_packet): Likewise.
13064         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13065         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13066         (final_emit_insn_group_barriers): Likewise.
13067         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13068         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13069         (mips16_insn_length): Likewise.
13070         * config/pa/pa.c (pa_reorg): Likewise.
13071         (pa_combine_instructions): Likewise.
13072         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13073         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13074         (sh_reorg): Likewise.
13075         (split_branches): Likewise.
13076         * config/spu/spu.c (get_branch_target): Likewise.
13078         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13079         JUMP_TABLE_DATA_P.
13081 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13083         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13084         Fix declaration name.
13086 2013-03-28  Lawrence Crowl  <crowl@google.com>
13088         * graphds.h (struct graph.indicies): Remove unused.
13089         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13090         (SCOP_ORIGINAL_PDDRS): Remove unused.
13091         * sese.h (extern insert_loop_close_phis): Removed unused.
13092         (extern insert_guard_phis): Removed unused.
13093         (extern ivtype_map_elt_info): Removed unused.
13094         (new_ivtype_map_elt): Removed unused.
13095         * sese.c (ivtype_map_elt_info): Removed unused.
13097 2013-03-28  Lawrence Crowl  <crowl@google.com>
13099         * Makefile.in: Add several missing include dependences.
13100         (DUMPFILE_H): New.
13101         (test-dump.o): New.  This object is not added to any executable,
13102         but is present for ad-hoc testing.
13103         * bitmap.c
13104         (debug (const bitmap_head_def &)): New.
13105         (debug (const bitmap_head_def *)): New.
13106         * bitmap.h
13107         (extern debug (const bitmap_head_def &)): New.
13108         (extern debug (const bitmap_head_def *)): New.
13109         * cfg.c
13110         (debug (edge_def &)): New.
13111         (debug (edge_def *)): New.
13112         * cfghooks.c
13113         (debug (basic_block_def &)): New.
13114         (debug (basic_block_def *)): New.
13115         * dumpfile.h
13116         (dump_node (const_tree, int, FILE *)): Correct source file.
13117         * dwarf2out.c
13118         (debug (die_struct &)): New.
13119         (debug (die_struct *)): New.
13120         * dwarf2out.h
13121         (extern debug (die_struct &)): New.
13122         (extern debug (die_struct *)): New.
13123         * gimple-pretty-print.c
13124         (debug (gimple_statement_d &)): New.
13125         (debug (gimple_statement_d *)): New.
13126         * gimple-pretty-print.h
13127         (extern debug (gimple_statement_d &)): New.
13128         (extern debug (gimple_statement_d *)): New.
13129         * ira-build.c
13130         (debug (ira_allocno_copy &)): New.
13131         (debug (ira_allocno_copy *)): New.
13132         (debug (ira_allocno &)): New.
13133         (debug (ira_allocno *)): New.
13134         * ira-int.h
13135         (extern debug (ira_allocno_copy &)): New.
13136         (extern debug (ira_allocno_copy *)): New.
13137         (extern debug (ira_allocno &)): New.
13138         (extern debug (ira_allocno *)): New.
13139         * ira-lives.c
13140         (debug (live_range &)): New.
13141         (debug (live_range *)): New.
13142         * lra-int.h
13143         (debug (lra_live_range &)): New.
13144         (debug (lra_live_range *)): New.
13145         * lra-lives.c
13146         (debug (lra_live_range &)): New.
13147         (debug (lra_live_range *)): New.
13148         * omega.c
13149         (debug (omega_pb_d &)): New.
13150         (debug (omega_pb_d *)): New.
13151         * omega.h
13152         (extern debug (omega_pb_d &)): New.
13153         (extern debug (omega_pb_d *)): New.
13154         * print-rtl.c
13155         (debug (const rtx_def &)): New.
13156         (debug (const rtx_def *)): New.
13157         * print-tree.c
13158         (debug_tree (tree): Move within file.
13159         (debug_raw (const tree_node &)): New.
13160         (debug_raw (const tree_node *)): New.
13161         (dump_tree_via_hooks (const tree_node *, int)): New.
13162         (debug (const tree_node &)): New.
13163         (debug (const tree_node *)): New.
13164         (debug_verbose (const tree_node &)): New.
13165         (debug_verbose (const tree_node *)): New.
13166         (debug_head (const tree_node &)): New.
13167         (debug_head (const tree_node *)): New.
13168         (debug_body (const tree_node &)): New.
13169         (debug_body (const tree_node *)): New.
13170         (debug_vec_tree (tree): Move and reimplement in terms of dump.
13171         (debug (vec<tree, va_gc> &)): New.
13172         (debug (vec<tree, va_gc> *)): New.
13173         * rtl.h
13174         (extern debug (const rtx_def &)): New.
13175         (extern debug (const rtx_def *)): New.
13176         * sbitmap.c
13177         (debug_raw (simple_bitmap_def &)): New.
13178         (debug_raw (simple_bitmap_def *)): New.
13179         (debug (simple_bitmap_def &)): New.
13180         (debug (simple_bitmap_def *)): New.
13181         * sbitmap.h
13182         (extern debug (simple_bitmap_def &)): New.
13183         (extern debug (simple_bitmap_def *)): New.
13184         (extern debug_raw (simple_bitmap_def &)): New.
13185         (extern debug_raw (simple_bitmap_def *)): New.
13186         * sel-sched-dump.c
13187         (debug (vinsn_def &)): New.
13188         (debug (vinsn_def *)): New.
13189         (debug_verbose (vinsn_def &)): New.
13190         (debug_verbose (vinsn_def *)): New.
13191         (debug (expr_def &)): New.
13192         (debug (expr_def *)): New.
13193         (debug_verbose (expr_def &)): New.
13194         (debug_verbose (expr_def *)): New.
13195         (debug (vec<rtx> &)): New.
13196         (debug (vec<rtx> *)): New.
13197         * sel-sched-dump.h
13198         (extern debug (vinsn_def &)): New.
13199         (extern debug (vinsn_def *)): New.
13200         (extern debug_verbose (vinsn_def &)): New.
13201         (extern debug_verbose (vinsn_def *)): New.
13202         (extern debug (expr_def &)): New.
13203         (extern debug (expr_def *)): New.
13204         (extern debug_verbose (expr_def &)): New.
13205         (extern debug_verbose (expr_def *)): New.
13206         (extern debug (vec<rtx> &)): New.
13207         (extern debug (vec<rtx> *)): New.
13208         * sel-sched-ir.h
13209         (_list_iter_cond_expr): Make inline instead of static.
13210         * sreal.c
13211         (debug (sreal &)): New.
13212         (debug (sreal *)): New.
13213         * sreal.h
13214         (extern debug (sreal &)): New.
13215         (extern debug (sreal *)): New.
13216         * tree.h
13217         (extern debug_raw (const tree_node &)): New.
13218         (extern debug_raw (const tree_node *)): New.
13219         (extern debug (const tree_node &)): New.
13220         (extern debug (const tree_node *)): New.
13221         (extern debug_verbose (const tree_node &)): New.
13222         (extern debug_verbose (const tree_node *)): New.
13223         (extern debug_head (const tree_node &)): New.
13224         (extern debug_head (const tree_node *)): New.
13225         (extern debug_body (const tree_node &)): New.
13226         (extern debug_body (const tree_node *)): New.
13227         (extern debug (vec<tree, va_gc> &)): New.
13228         (extern debug (vec<tree, va_gc> *)): New.
13229         * tree-cfg.c
13230         (debug (struct loop &)): New.
13231         (debug (struct loop *)): New.
13232         (debug_verbose (struct loop &)): New.
13233         (debug_verbose (struct loop *)): New.
13234         * tree-dump.c: Add header dependence.
13235         * tree-flow.h
13236         (extern debug (struct loop &)): New.
13237         (extern debug (struct loop *)): New.
13238         (extern debug_verbose (struct loop &)): New.
13239         (extern debug_verbose (struct loop *)): New.
13240         * tree-data-ref.c
13241         (debug (data_reference &)): New.
13242         (debug (data_reference *)): New.
13243         (debug (vec<data_reference_p> &)): New.
13244         (debug (vec<data_reference_p> *)): New.
13245         (debug (vec<ddr_p> &)): New.
13246         (debug (vec<ddr_p> *)): New.
13247         * tree-data-ref.h
13248         (extern debug (data_reference &)): New.
13249         (extern debug (data_reference *)): New.
13250         (extern debug (vec<data_reference_p> &)): New.
13251         (extern debug (vec<data_reference_p> *)): New.
13252         (extern debug (vec<ddr_p> &)): New.
13253         (extern debug (vec<ddr_p> *)): New.
13254         * tree-ssa-alias.c
13255         (debug (pt_solution &)): New.
13256         (debug (pt_solution *)): New.
13257         * tree-ssa-alias.h
13258         (extern debug (pt_solution &)): New.
13259         (extern debug (pt_solution *)): New.
13260         * tree-ssa-alias.c
13261         (debug (_var_map &)): New.
13262         (debug (_var_map *)): New.
13263         (debug (tree_live_info_d &)): New.
13264         (debug (tree_live_info_d *)): New.
13265         * tree-ssa-alias.h
13266         (extern debug (_var_map &)): New.
13267         (extern debug (_var_map *)): New.
13268         (extern debug (tree_live_info_d &)): New.
13269         (extern debug (tree_live_info_d *)): New.
13271 2013-03-28  Jan Hubicka  <jh@suse.cz>
13273         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13275 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
13277         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13278         record only when desired or required.
13280 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
13282         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13283         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
13285 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13287         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13288         (*andsi3_compare0_uxtw): New pattern.
13289         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13290         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13292 2013-03-28  Jan Hubicka  <jh@suse.cz>
13294         * data-streamer-in.c (streamer_read_gcov_count): New function.
13295         * gimple-streamer-out.c: Include value-prof.h.
13296         (output_gimple_stmt): Output histogram.
13297         (output_bb): Use streamer_write_gcov_count.
13298         * value-prof.c: Include data-streamer.h
13299         (dump_histogram_value): Add HIST_TYPE_MAX.
13300         (stream_out_histogram_value): New function.
13301         (stream_in_histogram_value): New function.
13302         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
13303         (stream_out_histogram_value, stream_in_histogram_value): Declare.
13304         * data-streamer-out.c (streamer_write_gcov_count): New function.
13305         (streamer_write_gcov_count_stream): New function.
13306         * lto-cgraph.c (lto_output_edge): Update counter streaming.
13307         (lto_output_node): Likewise.
13308         (input_node, input_edge): Likewise.
13309         * lto-streamer-out.c (output_cfg): Update streaming.
13310         * lto-streamer-in.c (input_cfg): Likewise.
13311         * data-streamer.h (streamer_write_gcov_count,
13312         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
13313         * gimple-streamer-in.c: Include value-prof.h
13314         (input_gimple_stmt): Input histograms.
13315         (input_bb): Update profile streaming.
13317 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13319         * genmodes.c (emit_max_int): New function.
13320         (emit_insn_modes_h): Added call to emit_max_function.
13321         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
13322         Added doc.
13323         * machmode.def: Fixed comment.
13325 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13327         * combine.c (try_combine): Removed useless assert.
13328         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
13330 2013-03-28  Marek Polacek  <polacek@redhat.com>
13331             Richard Biener  <rguenther@suse.de>
13333         PR tree-optimization/56695
13334         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
13335         build signed result of a vector comparison.
13336         * tree-cfg.c (verify_gimple_comparison): Check that a result
13337         of a vector comparison has signed type.
13339 2013-03-28  Richard Biener  <rguenther@suse.de>
13341         PR tree-optimization/37021
13342         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
13343         do not restrict gaps between groups.
13344         * tree-vect-stmts.c (vectorizable_load): Properly account for
13345         a gap between groups.
13347 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
13349         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
13350         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
13351         is not enabled.
13353 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
13355         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
13356         * doc/extend.texi (Named Address Spaces): Ditto.
13357         (Variable Attributes): Ditto.
13359 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13361         * config.build: Add support for cygwin x64 target.
13362         * config.gcc: Likewise.
13363         * config.host: Likewise.
13364         * configure.ac: Likewise
13365         * configure: Regenerated.
13367 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13369         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
13370         * config/i386/t-cygwin-w64: New file.
13371         * config/i386/cygwin-w64.h: New file.
13372         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
13373         and add support for x64-cygwin target.
13374         (CPP_SPEC): Likewise.
13375         (CXX_WRAP_SPEC_LIST): Undefine before define.
13376         (LIBGCJ_SONAME): Use 15 as version.
13378 2013-03-27  Richard Biener  <rguenther@suse.de>
13380         PR tree-optimization/56716
13381         * tree-ssa-structalias.c (perform_var_substitution): Adjust
13382         dumping for ref nodes.
13384 2013-03-27  Martin Jambor  <mjambor@suse.cz>
13386         PR tree-optimization/55334
13387         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
13388         restricted pointers to arrays.
13390 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13392         * Makefile.in (.SUFFIXES): Add .cc.
13393         (.c.o): Apply same recipe for implicit rule .cc.o.
13395 2013-03-27  Richard Biener  <rguenther@suse.de>
13397         PR tree-optimization/37021
13398         * tree-vect-data-refs.c (vect_check_strided_load): Allow
13399         REALPART/IMAGPART_EXPRs around the supported refs.
13400         * tree-ssa-structalias.c (find_func_aliases): Assume that
13401         floating-point values are not used to transfer pointers.
13403 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13405         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
13406         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
13407         * gcc/doc/tm.texi: Regenerate.
13408         * gcc/targhooks.h (default_has_ifunc_p): New.
13409         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
13410         * gcc/config/linux-protos.h: New file.
13411         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
13412         this hook for linux which disables support of indirect functions in
13413         android.
13414         * gcc/config/linux-android.c: New file.
13415         * gcc/config/t-linux-android.c: Ditto.
13416         * gcc/config.gcc: Added new object file linux-android.o.
13417         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
13418         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
13419         * gcc/varasm.c (do_assemble_alias): Likewise.
13420         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
13421         doesn't support indirect functions.
13422         * configure: Regenerate.
13424 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
13426         PR target/56102
13427         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
13428         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
13429         mult-word mode.
13431 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13433         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
13435 2013-03-27  Terry Guo  <terry.guo@arm.com>
13437         * config/arm/arm-cores.def: Added core cortex-r7.
13438         * config/arm/arm-tune.md: Regenerated.
13439         * config/arm/arm-tables.opt: Regenerated.
13440         * doc/invoke.texi: Added entry for core cortex-r7.
13442 2013-03-27  Walter Lee  <walt@tilera.com>
13444         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
13445         double-decrement of next_scratch_regno.
13447 2013-03-27  Walter Lee  <walt@tilera.com>
13449         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
13450         input operands.
13451         (insn_v1mulus): Ditto.
13452         (insn_v2muls): Ditto.
13454 2013-03-27  Walter Lee  <walt@tilera.com>
13456         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
13457         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
13459 2013-03-27  Walter Lee  <walt@tilera.com>
13461         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
13462         (*sibcall_value): Ditto.
13464 2013-03-27  Walter Lee  <walt@tilera.com>
13466         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
13467         (insn_mnz_v8qi): ... this ...
13468         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
13469         vector equivalent.
13470         (insn_v<n>mnz): Replaced by ...
13471         (insn_v1mnz): ... this ...
13472         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
13473         equivalent.
13474         (insn_mz_<mode>): Replaced by ...
13475         (insn_mz_v8qi): ... this ...
13476         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
13477         vector equivalent.
13478         (insn_v<n>mz): Replaced by ...
13479         (insn_v1mz): ... this ...
13480         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
13481         equivalent.
13483 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
13485         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
13487 2013-03-26  Roland McGrath  <mcgrathr@google.com>
13489         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
13490         than fprintf with a non-constant, non-format string.
13492 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13494         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
13495         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
13496         operand 0 predicate.
13497         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
13498         attribute.  Use general_x64nomem_operand as operand 1 predicate.
13499         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
13500         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13501         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
13502         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13503         (mov<mode>_insv_1): Remove expander.  Merge insn with
13504         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
13505         Use general_x64nomem_operand as operand 1 predicate.
13506         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
13507         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
13508         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13509         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
13510         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
13511         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
13512         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13513         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
13514         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
13515         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
13516         (general_x64nomem_operand): Ditto.
13518 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13520         * config/rtems.opt: Add -pthread option.
13522 2013-03-26  Richard Biener  <rguenther@suse.de>
13524         * alias.c (find_base_term): Avoid redundant and not used recursion.
13525         (base_alias_check): Get the initial base term from the caller.
13526         (true_dependence_1): Compute and pass base terms to base_alias_check.
13527         (write_dependence_p): Likewise.
13528         (may_alias_p): Likewise.
13530 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
13532         * config/aarch64/aarch64.c (aarch64_classify_address): Support
13533         PC-relative load in SI modes and above only.
13535 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
13537         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
13538         * config/arm/iwmmxt.md (WCGR0): Update.
13539         (WCGR1, WCGR2, WCGR3): Likewise.
13541 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13543         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13544         Use x64 and nox64 isa attributes.
13546 2013-03-26  Richard Biener  <rguenther@suse.de>
13548         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13549         alignment computations and rely on get_object_alignment_1
13550         for the !TYPE_P case.
13551         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13553 2013-03-26  Walter Lee  <walt@tilera.com>
13555         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13556         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13558 2013-03-25  Jeff Law  <law@redhat.com>
13560         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13561         check for INTEGRAL_TYPE_P that was missing due to checking in
13562         wrong version of prior patch.
13564 2013-03-25  Walter Lee  <walt@tilera.com>
13566         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13567         TILEGX_INSN_SHUFFLEBYTES1.
13568         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13569         shufflebytes1.
13570         (tilegx_builtins): Ditto.
13571         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13573 2013-03-25  Walter Lee  <walt@tilera.com>
13575         * config/tilegx/tilegx.md (floatsisf2): New pattern.
13576         (floatunssisf2): New pattern.
13577         (floatsidf2): New pattern.
13578         (floatunssidf2): New pattern.
13580 2013-03-25  Walter Lee  <walt@tilera.com>
13582         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13583         tests for constraint J, K, N, P.
13585 2013-03-25  Walter Lee  <walt@tilera.com>
13587         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13588         Use indirect/pcrel encoding.
13589         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13590         Ditto.
13592 2013-03-25  Steve Ellcey  <sellcey@mips.com>
13594         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13595         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13596         * config/mips/mips.c (mips_option_override): Set IMADD default.
13597         * config/mips/mips.h (PTF_AVOID_IMADD): New.
13598         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13599         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13600         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13601         * doc/invoke.texi (-mimadd/-mno-imadd): New.
13603 2013-03-25  Jeff Law  <law@redhat.com>
13605         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13606         slightly to avoid creating and folding useless trees.  Simplify
13607         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13609 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
13611         * config/i386/i386.md (*zero_extendsidi2): Merge with
13612         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
13613         * config/i386/predicates.md (x86_64_zext_operand): Rename from
13614         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
13615         targets.  Clarify comment.
13617 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13619         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13620         pass-through jump functions differently.
13621         (ipa_read_jump_function): Likewise.  Also use setter functions to set
13622         up jump functions.
13624 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13626         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13627         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13628         process it.
13629         (ipa_get_indirect_edge_target): New function.
13630         (devirtualization_time_bonus): New parameter known_aggs, pass it to
13631         ipa_get_indirect_edge_target.  Update all callers.
13632         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
13633         ipa_get_indirect_edge_target_1 instead of calling
13634         ipa_get_indirect_edge_target.
13635         (create_specialized_node): Pass aggvlas to
13636         ipcp_discover_new_direct_edges.
13638 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13640         * config/arm/arm.md (f_sels, f_seld): New types.
13641         (*cmov<mode>): New pattern.
13642         * config/arm/predicates.md (arm_vsel_comparison_operator): New
13643         predicate.
13645 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13647         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13648         POSIX-printf for mingw-hosted  builds.
13650 2013-03-25  Richard Biener  <rguenther@suse.de>
13652         PR middle-end/56694
13653         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13654         must-not-throw stmt location.
13656 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13658         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13659         Emit load-acquire versions when acq is true.
13660         (arm_emit_store_exclusive): Add rel parameter.
13661         Emit store-release versions when rel is true.
13662         (arm_split_compare_and_swap): Use acquire-release instructions
13663         instead.
13664         of barriers when appropriate.
13665         (arm_split_atomic_op): Likewise.
13666         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13667         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13668         (VUNSPEC_SLX): Likewise.
13669         (VUNSPEC_LDA): Likewise.
13670         (VUNSPEC_STL): Likewise.
13671         * config/arm/sync.md (atomic_load<mode>): New pattern.
13672         (atomic_store<mode>): Likewise.
13673         (arm_load_acquire_exclusive<mode>): Likewise.
13674         (arm_load_acquire_exclusivesi): Likewise.
13675         (arm_load_acquire_exclusivedi): Likewise.
13676         (arm_store_release_exclusive<mode>): Likewise.
13678 2013-03-25  Catherine Moore  <clm@codesourcery.com>
13680         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13681         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13682         * config/mip/predicates.md (lwsp_swsp_operand,
13683         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13684         sb16_operand, db4_operand, db7_operand, ib3_operand,
13685         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13686         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13687         andi16_operand): New predicates.
13688         * config/mips/mips.md (compression): New attribute.
13689         (enabled): New attribute.
13690         (length): Consider compression in computing length.
13691         (shift_compression): New code attribute.
13692         (*add<mode>3): New operands. Record compression.
13693         (sub<mode>3): Likewise.
13694         (one_cmpl<mode>2): Likewise.
13695         (*and<mode>3): Likewise.
13696         (*ior<mode>3): Likewise.
13697         (unnamed pattern for xor): Likewise.
13698         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13699         (*<optab><mode>3): Likewise.
13700         (*mov<mode>_internal: Likewise.
13701         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13702         (mips_unsigned_immediate_p): New.
13703         (umips_lwsp_swsp_address_p): New.
13704         (m16_based_address_p): New.
13705         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13706         (mips_unsigned_immediate_p): New prototype.
13707         (lwsp_swsp_address_p): New prototype.
13708         (m16_based_address_p): New prototype.
13709         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
13710         (mips_signed_immediate_p): New function.
13711         (m16_based_address_p): New function.
13712         (lwsp_swsp_address_p): New function.
13713         (mips_print_operand_punctuation): Recognize short delay slot insns
13714         for microMIPS.add<mode>3"
13716 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13718         PR target/56720
13719         * config/arm/iterators.md (v_cmp_result): New mode attribute.
13720         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
13722 2013-03-25  Richard Biener  <rguenther@suse.de>
13724         PR tree-optimization/56689
13725         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
13726         any edge.
13728 2013-03-25  Richard Biener  <rguenther@suse.de>
13730         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
13731         of bitmap.
13732         (memory_references): Likewise.
13733         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
13734         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
13735         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
13736         (gather_mem_refs_in_loops): Fold into ...
13737         (analyze_memory_references): ... this.  Move initialization
13738         to tree_ssa_lim_initialize.
13739         (fill_always_executed_in): Rename to ...
13740         (fill_always_executed_in_1): ... this.
13741         (fill_always_executed_in): Move contains_call computation to
13742         this new function from ...
13743         (tree_ssa_lim_initialize): ... here.
13744         (tree_ssa_lim): Call fill_always_executed_in.
13746 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13748         * postreload.c (reload_combine): Fix code detecting returns.
13750 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13752         * function.c (emit_use_return_register_into_block): On cc0 targets,
13753         do not emit the sequence between cc0 setter and user.
13755 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13757         * config/i386/predicates.md (local_symbolic_operand): Interpret
13758         dll-imported symbols as none-local.
13760 2013-03-25  Richard Biener  <rguenther@suse.de>
13762         * tree-ssa-loop-im.c (struct depend): Remove.
13763         (struct lim_aux_data): Make depends a vec of gimples.
13764         (free_lim_aux_data): Adjust.
13765         (add_dependency): Likewise.
13766         (set_level): Likewise.
13768 2013-03-25  Richard Biener  <rguenther@suse.de>
13770         PR middle-end/56434
13771         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
13772         the pointer returned by calls with ECF_MALLOC set.
13774 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13776         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
13778 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13780         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
13781         using MMXMODE mode iterator.
13782         (*move<mode>_internal): Merge with *movv2sf_internal and
13783         *movv2sf_internal_rex64 using MMXMODE mode iterator.
13785 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13787         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
13788         (record_last_mem_set_info): Likewise.
13790         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
13791         of XNEWVEC followed by memset.
13792         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
13794 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13796         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
13797         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
13798         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
13799         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
13800         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
13801         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13802         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13803         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13804         BARRIER_P instead of GET_CODE.
13806 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13808         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13809         inaccuracy in the probing code.
13811         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13812         (ctrapdi4): Likewise.
13814 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13816         * calls.c (expand_call): Add missing guard to code handling return
13817         of non-BLKmode structures in MSB.
13818         * function.c (expand_function_end): Likewise.
13820 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13822         * combine.c (try_combine): Adjust comment.  Do not add the set of
13823         insn #0 if the destination indirectly is set or dies in insn #2.
13824         Tidy up code to distribute a new note.
13826 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13828         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13829         also for alternatives 16 and 17.
13831 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13833         * config/i386/sse.md (*mov<mode>_internal): Merge with
13834         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
13835         Emit insn template depending on type attribute.  Use
13836         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13837         movd instead of movq mnemonic for interunit moves.  Rewrite mode
13838         attribute calculation.  Remove unit attribute calculation.
13839         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13840         Set prefix_data16 attribute for DImode ssemov types.
13841         Use Ym instead of y for SSE-MMX conversion alternatives.
13842         Reorder operand constraints.
13844 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
13846         * df.h (df_insn_delete): Adjust prototype.
13847         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13848         and let it decide whether mark the basic block dirty.
13849         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13850         * df-scan.c (df_insn_info_delete): New helper function, split
13851         off from df_insn_delete.
13852         (df_scan_free_bb_info): Use it.
13853         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13854         Likewise.
13855         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
13856         that the insn is actually an insn and it has a non-NULL basic block.
13857         Do not mark basic block dirty if only deleting a DEBUG_INSN.
13859 2013-03-22  Richard Biener  <rguenther@suse.de>
13861         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13862         dep_ref members.
13863         (mem_ref_alloc): Do not allocate them.
13864         (refs_independent_p): Do not query or maintain a cache.
13866 2013-03-22  Richard Biener  <rguenther@suse.de>
13868         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13869         (gather_mem_refs_in_loops): Do not compute it.
13870         (analyze_memory_references): Do not allocate it.
13871         (tree_ssa_lim_finalize): Do not free it.
13872         (for_all_locs_in_loop): Do not query all_refs_in_loop.
13874 2013-03-22  Richard Biener  <rguenther@suse.de>
13876         * is-a.h (as_a): Use gcc_checking_assert.
13878 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
13880         * config/aarch64/aarch64.c (aarch64_print_operand): New
13881         format specifier for printing a constant in hex.
13882         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13883         format specifier for printing second operand.
13885 2013-03-22  Richard Biener  <rguenther@suse.de>
13887         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13888         bitmaps.
13889         (gather_mem_refs_in_loops): Perform store accumulation here.
13890         (create_vop_ref_mapping_loop): Remove.
13891         (create_vop_ref_mapping): Likewise.
13892         (analyze_memory_references): Initialize refs_stored_in_loop.
13893         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13894         (record_indep_loop): Remove.
13895         (record_dep_loop): New function.
13896         (ref_indep_loop_p_1): Adjust to only walk over references
13897         in the loop, not its subloops.
13898         (ref_indep_loop_p): Rename to ...
13899         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13900         maintaining a more fine-grained cache.
13901         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13902         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13904 2013-03-22  Richard Biener  <rguenther@suse.de>
13906         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13907         (struct mem_ref): Make accesses_in_loop a vec of a vec of
13908         aggregate mem_ref_loc.
13909         (free_mem_ref_locs): Inline into ...
13910         (memref_free): ... this and adjust.
13911         (mem_ref_alloc): Adjust.
13912         (mem_ref_locs_alloc): Remove.
13913         (record_mem_ref_loc): Adjust.
13914         (get_all_locs_in_loop): Rewrite into ...
13915         (for_all_locs_in_loop): ... this iterator.
13916         (rewrite_mem_ref_loc): New functor.
13917         (rewrite_mem_refs): Use for_all_locs_in_loop.
13918         (sm_set_flag_if_changed): New functor.
13919         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13920         (ref_always_accessed): New functor.
13921         (ref_always_accessed_p): Use for_all_locs_in_loop.
13923 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13925         * tree-pass.h (PROP_gimple_lvec): New.
13926         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13927         (init_optimization_passes): Move pass_lower_vector.
13928         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13929         PROP_gimple_lvec.
13930         (pass_lower_vector): Provide PROP_gimple_lvec.
13931         (pass_lower_vector_ssa): Likewise.
13932         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13934 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13936         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13938 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13940         * config/i386/i386.md (*movdi_internal): Disparage slightly
13941         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
13942         conversion alternatives.
13944 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
13946         PR middle-end/48087
13947         * diagnostic.def (DK_WERROR): New kind.
13948         * diagnostic.h (werrorcount): Define.
13949         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13950         promoted to DK_ERROR, increment DK_WERROR counter instead of
13951         DK_ERROR counter.
13952         * toplev.c (toplev_main): Call print_ignored_options even if
13953         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13954         even if just werrorcount is non-zero.
13956         PR debug/55608
13957         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13958         on failure.
13959         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13960         (string_cst_pool_decl): New function.
13961         (optimize_one_addr_into_implicit_ptr): New function.
13962         (resolve_addr_in_expr): Optimize DWARF location expression
13963         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13964         which doesn't live in memory, but has DW_AT_location or
13965         DW_AT_const_value, or refers to a string literal, into
13966         DW_OP_GNU_implicit_pointer.
13967         (optimize_location_into_implicit_ptr): New function.
13968         (resolve_addr): If removing DW_AT_location of a variable because
13969         it was DW_OP_addr of address of the variable, but the variable doesn't
13970         live in memory, try to emit const value attribute for the initializer.
13972 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13974         * tree.h (VECTOR_TYPE_P): New macro.
13975         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13976         TYPE_MODE): Use it.
13977         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13978         VEC_COND_EXPR cannot be lvalues.
13979         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13981 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13983         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13984         Restrict the transformation to equal modes.
13986 2013-03-21  Richard Biener  <rguenther@suse.de>
13988         PR tree-optimization/39326
13989         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13990         (MEM_ANALYZABLE): Adjust.
13991         (record_mem_ref_loc): Move bitmap ops ...
13992         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
13993         unanalyzable refs, do not record locations for it.
13994         (analyze_memory_references): Allocate ref zero as shared
13995         unanalyzable ref.
13996         (refs_independent_p): Do not test for unanalyzed mems here.
13997         (ref_indep_loop_p_1): Special-case disambiguation against
13998         the unanalyzed ref.
13999         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14001 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
14003         * config/arm/arm-protos.h (tune_params): Add
14004         prefer_neon_for_64bits field.
14005         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14006         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14007         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14008         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14009         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14010         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14011         (arm_option_override): Handle -mneon-for-64bits new option.
14012         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14013         (prefer_neon_for_64bits): Declare new variable.
14014         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14015         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14016         (arch_enabled): Handle new arch types. Remove support for onlya8
14017         and nota8.
14018         (one_cmpldi2): Use new arch names.
14019         (zero_extend<mode>di2, extend<mode>di2): Ditto.
14020         * config/arm/arm.opt (mneon-for-64bits): Add option.
14021         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14022         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14023         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14024         of onlya8.
14025         * doc/invoke.texi (-mneon-for-64bits): Document.
14027 2013-03-21  Richard Biener  <rguenther@suse.de>
14029         PR tree-optimization/39326
14030         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14031         (sort_bbs_in_loop_postorder_cmp): New function.
14032         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14034 2013-03-21  Richard Biener  <rguenther@suse.de>
14036         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14037         (vect_insert_into_interleaving_chain): Likewise.
14038         (vect_drs_dependent_in_basic_block): Inline ...
14039         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
14040         split out from ...
14041         (vect_analyze_data_ref_dependence): ... here.  Simplify.
14042         (vect_check_interleaving): Simplify.
14043         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
14044         (vect_slp_analyze_data_ref_dependences): ... this new function.
14045         (dr_group_sort_cmp): New function.
14046         (vect_analyze_data_ref_accesses): Compute data-reference groups
14047         here instead of in vect_analyze_data_ref_dependence.  Use
14048         a more efficient algorithm.
14049         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14050         vect_slp_analyze_data_ref_dependences.  Call
14051         vect_analyze_data_ref_accesses earlier.
14052         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14053         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14054         (vect_slp_analyze_data_ref_dependences): New prototype.
14056 2013-03-21  Richard Biener  <rguenther@suse.de>
14058         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14059         ref is stored in the loop.
14060         (find_refs_for_sm): Walk only over all stores.
14061         (store_motion_loop): Allocate from lim_bitmap_obstack.
14062         (store_motion): Likewise.
14064 2013-03-21  Richard Biener  <rguenther@suse.de>
14066         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14067         Update virtual SSA form.
14069 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14071         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14072         * configure: Regenerate.
14073         * config.in: Regenerate.
14074         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14075         if !HAVE_LD_EH_FRAME_CIEV3.
14077 2013-03-21  Richard Biener  <rguenther@suse.de>
14079         * tree-cfg.c (verify_expr_no_block): New function.
14080         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14081         nor DECL_VALUE_EXPR have locations with associated blocks.
14082         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14083         (clear_unused_block_pointer): Remove code dealing with
14084         blocks in DECL_DEBUG_EXPR locations.
14086 2013-03-21  Richard Biener  <rguenther@suse.de>
14088         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14089         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14090         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14091         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14092         instead of DECL_DEBUG_EXPR_IS_FROM.
14093         * gimplify.c (gimplify_modify_expr): Likewise.
14094         * tree-cfg.c (verify_expr_location_1): Likewise.
14095         * tree-complex.c (create_one_component_var): Likewise.
14096         * tree-sra.c (create_access_replacement): Likewise.
14097         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14098         (clear_unused_block_pointer): Likewise.
14099         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14100         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14101         * var-tracking.c (var_debug_decl): Likewise.
14102         (track_expr_p): Likewise.
14103         * tree-inline.c (add_local_variables): Likewise.  Set
14104         DECL_HAS_DEBUG_EXPR_P after copying it.
14105         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14106         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
14108 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14110         PR bootstrap/56656
14111         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14112         * configure: Regenerate.
14113         * config.in: Regenerate.
14114         * config/i386/i386.md (*movdf_internal): Use
14115         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14116         movd instead of movq mnemonic for interunit moves.
14117         (*movdi_internal): Ditto.
14119 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14121         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14122         (abd<mode>_3): New pattern.
14123         (aba<mode>_3): New pattern.
14124         (fabd<mode>_3): New pattern.
14126 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14128         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14129         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14130         occurrence of REGISTER_PREFIX as its empty string.
14132 2013-03-20  Jeff Law  <law@redhat.com>
14134         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14135         addititional equivalences for equality comparisons between an SSA_NAME
14136         and a constant where the SSA_NAME was set from a widening conversion.
14138 2013-03-20  Walter Lee  <walt@tilera.com>
14140         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14142 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14144         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14145         depending on type attribute.
14146         (*movti_internal): Ditto.
14147         (*movtf_internal): Ditto.
14148         (*movxf_internal): Ditto.
14149         (*movdf_internal): Ditto.
14150         (*movsf_internal): Ditto.
14152 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14154         * config/i386/i386.md (*movti_internal): Set prefix attribute to
14155         maybe_vex for sselog1 and ssemov types.
14156         (*movdi_internal): Reorder operand constraints.
14157         (*movsi_internal): Ditto.  Set prefix attribute to
14158         maybe_vex for sselog1 and ssemov types.
14159         (*movtf_internal): Set prefix attribute to maybe_vex
14160         for sselog1 and ssemov types.
14161         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
14162         DImode ssemov types.  Reorder operand constraints.
14163         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
14164         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
14165         attribute for SImode ssemov types.  Reorder operand constraints.
14167 2013-03-20  Martin Jambor  <mjambor@suse.cz>
14169         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14170         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14172 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
14174         * config/rs6000/predicates.md (indexed_address, update_address_mem
14175         update_indexed_address_mem): New predicates.
14176         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14177         attribute for load/store instructions.
14178         * config/rs6000/dfp.md (movsd_store): Likewise.
14179         (movsd_load): Likewise.
14180         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14181         (unnamed HI->DI extend define_insn): Likewise.
14182         (unnamed SI->DI extend define_insn): Likewise.
14183         (unnamed QI->SI extend define_insn): Likewise.
14184         (unnamed QI->HI extend define_insn): Likewise.
14185         (unnamed HI->SI extend define_insn): Likewise.
14186         (unnamed HI->SI extend define_insn): Likewise.
14187         (extendsfdf2_fpr): Likewise.
14188         (movsi_internal1): Likewise.
14189         (movsi_internal1_single): Likewise.
14190         (movhi_internal): Likewise.
14191         (movqi_internal): Likewise.
14192         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14193         attribute for load/store instructions.
14194         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14195         instructions.
14196         (mov<mode>_softfloat): Likewise.
14197         (mov<mode>_hardfloat32): Likewise.
14198         (mov<mode>_hardfloat64): Likewise.
14199         (mov<mode>_softfloat64): Likewise.
14200         (movdi_internal32): Likewise.
14201         (movdi_internal64): Likewise.
14202         (probe_stack_<mode>): Likewise.
14204 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
14206         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14207         floating point, and decimal floating point to reload iterator.
14209         * config/rs6000/constraints.md (wl constraint): New constraints to
14210         return FLOAT_REGS if certain options are used to reduce the number
14211         of separate patterns that exist in the file.
14212         (wx constraint): Likewise.
14213         (wz constraint): Likewise.
14215         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14216         -mdebug=reg, print wg, wl, wx, and wz constraints.
14217         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14218         Initialize the reload functions for 64-bit binary/decimal floating
14219         point types.
14220         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14221         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14222         create the buffer on the stack to overcome not having a 32-bit
14223         load and store.
14224         (rs6000_emit_move): Likewise.
14225         (rs6000_secondary_memory_needed_rtx): Likewise.
14226         (rs6000_alloc_sdmode_stack_slot): Likewise.
14227         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14228         via xxlxor, just like DFmode 0.0.
14230         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14231         define as 1 if we are running on a power7 or newer.
14232         (enum r6000_reg_class_enum): Add new constraints.
14234         * config/rs6000/dfp.md (movsd): Delete, combine with binary
14235         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14236         with other moves by using conditional constraits (wg).  Use LFIWZX
14237         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
14238         (movsd splitter): Likewise.
14239         (movsd_hardfloat): Likewise.
14240         (movsd_softfloat): Likewise.
14242         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14243         binary and decimal floating point moves.
14244         (fmove_ok): New attributes to combine binary and decimal floating
14245         point moves, and to combine power6x (mfpgpr) moves along normal
14246         floating moves.
14247         (real_value_to_target): Likewise.
14248         (f32_lr): Likewise.
14249         (f32_lm): Likewise.
14250         (f32_li): Likewise.
14251         (f32_sr): Likewise.
14252         (f32_sm): Likewise.
14253         (f32_si): Likewise.
14254         (movsf): Combine binary and decimal floating point moves.  Combine
14255         power6x (mfpgpr) moves with other moves by using conditional
14256         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
14257         (mov<mode> for SFmode/SDmode); Likewise.
14258         (SFmode/SDmode splitters): Likewise.
14259         (movsf_hardfloat): Likewise.
14260         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14261         (movsf_softfloat): Likewise.
14262         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14264         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14265         wx and wz constraints.
14267         * config/rs6000/constraints.md (wg constraint): New constraint to
14268         return FLOAT_REGS if -mmfpgpr (power6x) was used.
14270         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14271         constraint.
14273         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14274         -mdebug=reg, print wg, wl, wx, and wz constraints.
14275         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14276         Initialize the reload functions for 64-bit binary/decimal floating
14277         point types.
14278         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14279         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14280         create the buffer on the stack to overcome not having a 32-bit
14281         load and store.
14282         (rs6000_emit_move): Likewise.
14283         (rs6000_secondary_memory_needed_rtx): Likewise.
14284         (rs6000_alloc_sdmode_stack_slot): Likewise.
14285         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14286         via xxlxor, just like DFmode 0.0.
14288         * config/rs6000/dfp.md (movdd): Delete, combine with binary
14289         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14290         with other moves by using conditional constraits (wg).  Use LFIWZX
14291         and STFIWX for loading SDmode on power7.
14292         (movdd splitters): Likewise.
14293         (movdd_hardfloat32): Likewise.
14294         (movdd_softfloat32): Likewise.
14295         (movdd_hardfloat64_mfpgpr): Likewise.
14296         (movdd_hardfloat64): Likewise.
14297         (movdd_softfloat64): Likewise.
14299         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14300         64-bit binary and decimal floating point moves.
14301         (FMOVE64X): Likewise.
14302         (movdf): Combine 64-bit binary and decimal floating point moves.
14303         Combine power6x (mfpgpr) moves with other moves by using
14304         conditional constraits (wg).
14305         (mov<mode> for DFmode/DDmode): Likewise.
14306         (DFmode/DDmode splitters): Likewise.
14307         (movdf_hardfloat32): Likewise.
14308         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
14309         (movdf_softfloat32): Likewise.
14310         (movdf_hardfloat64_mfpgpr): Likewise.
14311         (movdf_hardfloat64): Likewise.
14312         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
14313         (movdf_softfloat64): Likewise.
14314         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
14315         (reload_<mode>_load): Move to later in the file so they aren't in
14316         the middle of the floating point move insns.
14317         (reload_<mode>_store): Likewise.
14319         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
14320         constraint.
14322         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
14323         constraint if -mdebug=reg.
14324         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
14325         Enable using dd reload support if needed.
14327         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
14328         binary and decimal floating point moves in rs6000.md.
14329         (movtd_internal): Likewise.
14331         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
14332         decimal floating point moves.
14333         (movtf): Likewise.
14334         (movtf_internal): Likewise.
14335         (mov<mode>_internal, TDmode/TFmode): Likewise.
14336         (movtf_softfloat): Likewise.
14337         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
14339         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
14340         movdi_internal64, using wg constraint for move direct operations.
14341         (movdi_internal64): Likewise.
14343         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
14344         MODES_TIEABLE_P for selected modes.  Print the numerical value of
14345         the various virtual registers. Use GPR/FPR first/last values,
14346         instead of hard coding the register numbers.  Print which modes
14347         have reload functions registered.
14348         (rs6000_option_override_internal): If -mdebug=reg, trace the options
14349         settings before/after setting cpu, target and subtarget settings.
14350         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
14351         and for secondary reload failures in rs6000_secondary_reload_inner.
14352         (rs6000_secondary_reload_fail): Likewise.
14353         (rs6000_secondary_reload_inner): Likewise.
14355         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
14356         macros for first/last GPR and FPR registers.
14357         (LAST_GPR_REGNO): Likewise.
14358         (FIRST_FPR_REGNO): Likewise.
14359         (LAST_FPR_REGNO): Likewise.
14361         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
14362         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
14363         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
14364         (vcond<mode><mode>): Likewise.
14365         (vcondu<mode><mode>): Likewise.
14366         (vector_gtu<mode>): Likewise.
14367         (vector_gte<mode>): Likewise.
14368         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
14369         to prevent the compiler from converting DImode operations to TImode.
14370         (ior<mode>3): Likewise.
14371         (and<mode>3): Likewise.
14372         (one_cmpl<mode>2): Likewise.
14373         (nor<mode>3): Likewise.
14374         (andc<mode>3): Likewise.
14376         * config/rs6000/constraints.md (wt constraint): New constraint
14377         that returns VSX_REGS if TImode is allowed in VSX registers.
14379         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
14380         constant under VSX.
14382         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
14383         similar to TImode, but it is restricted to being in the GPRs.
14385         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
14386         TImode to occupy a single VSX register.
14388         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
14389         -mvsx-timode for power7/power8.
14390         (power7 cpu): Likewise.
14391         (power8 cpu): Likewise.
14393         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
14394         sure that TFmode/TDmode take up two registers if they are ever
14395         allowed in the upper VSX registers.
14396         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
14397         registers.
14398         (rs6000_init_hard_regno_mode_ok): Likewise.
14399         (rs6000_debug_reg_global): Add debugging for PTImode and wt
14400         constraint.  Print if LRA is turned on.
14401         (rs6000_option_override_internal): Give an error if -mvsx-timode
14402         and VSX is not enabled.
14403         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
14404         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
14405         to reg+offset addressing.  Use PTImode when checking offset
14406         addresses for validity.
14407         (reg_offset_addressing_ok_p): Likewise.
14408         (rs6000_legitimate_offset_address_p): Likewise.
14409         (rs6000_legitimize_address): Likewise.
14410         (rs6000_legitimize_reload_address): Likewise.
14411         (rs6000_legitimate_address_p): Likewise.
14412         (rs6000_eliminate_indexed_memrefs): Likewise.
14413         (rs6000_emit_move): Likewise.
14414         (rs6000_secondary_reload): Likewise.
14415         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
14416         reloads to fpr registers to continue to use reg+offset addressing,
14417         but 64-bit reloads to altivec registers need reg+reg addressing.
14418         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
14419         it.  Treat LO_SUM like a PLUS operation.
14420         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
14421         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
14422         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
14423         registers to share a register with a smaller sized type, since VSX
14424         puts scalars in the upper 64-bits.
14425         (print_operand): Add support for PTImode.
14426         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
14427         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
14428         registers, but don't have arithmetic support.
14429         (rs6000_memory_move_cost): Add test for VSX.
14430         (rs6000_opt_masks): Add -mvsx-timode.
14432         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
14433         for TImode.
14434         (VSs): Likewise.
14435         (VSr): Use wt constraint for TImode.
14436         (VSv): Drop TImode support.
14437         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
14438         (vsx_movti_64bit): Likewise.
14439         (vsx_movti_32bit): Likewise.
14440         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
14441         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
14442         one '?' on the appropriate output constraint.  Do not allow TImode
14443         logical operations on 32-bit systems.
14444         (vsx_ior<mode>3): Likewise.
14445         (vsx_xor<mode>3): Likewise.
14446         (vsx_one_cmpl<mode>2): Likewise.
14447         (vsx_nor<mode>3): Likewise.
14448         (vsx_andc<mode>3): Likewise.
14449         (vsx_concat_<mode>): Likewise.
14450         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
14452         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
14453         OPTION_MASK_VSX_TIMODE.
14454         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
14455         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
14457         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
14458         (TI2 iterator): New iterator for TImode, PTImode.
14459         (wd mode attribute): Add values for vector types.
14460         (movti_string): Replace TI move operations with operations for TImode
14461         and PTImode.  Add support for TImode being allowed in VSX registers.
14462         (mov<mode>_string, TImode/PTImode): Likewise.
14463         (movti_ppc64): Likewise.
14464         (mov<mode>_ppc64, TImode/PTImode): Likewise.
14465         (TI mode splitters): Likewise.
14467         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
14468         constraint.
14470 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
14472         PR tree-optimization/56355
14473         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
14474         Also handle integers with undefined overflow.
14476 2013-03-20  Catherine Moore  <clm@codesourcery.com>
14477             Maciej W. Rozycki  <macro@codesourcery.com>
14478             Tom de Vries  <tom@codesourcery.com>
14479             Nathan Sidwell <nathan@codesourcery.com>
14480             Iain Sandoe  <iain@codesourcery.com>
14481             Nathan Froyd  <froydnj@codesourcery.com>
14482             Chao-ying Fu <fu@mips.com>
14484         * doc/extend.texi: (micromips, nomicromips, nocompression):
14485         Document new function attributes.
14486         * doc/invoke.texi (minterlink-compressed, mmicromips,
14487         m14k, m14ke, m14kec): Document new options.
14488         (minterlink-mips16): Update documentation.
14489         * doc/md.texi (ZC, ZD): Document new constraints.
14490         * configure.ac (gcc_cv_as_micromips): Check if linker
14491         supports the .set micromips directive.
14492         * configure: Regenerate.
14493         * config.in: Regenerate.
14494         * config/mips/mips-tables.opt: Regenerate.
14495         * config/mips/micromips.md: New file.
14496         * constraints.md (ZC, ZD): New constraints.
14497         * config/mips/predicates.md (movep_src_register): New predicate.
14498         (movep_src_operand): New predicate.
14499         (non_volatile_mem_operand): New predicate.
14500         * config/mips/mips.md (multimem): New type.
14501         (length): Differentiate between 17-bit and 18-bit branch offsets.
14502         (MOVEP1, MOVEP2): New mode iterator.
14503         (mov_<load>l): Use ZC constraint.
14504         (mov_<load>r): Likewise.
14505         (mov_<store>l): Likewise.
14506         (mov_<store>r): Likewise.
14507         (*branch_equality<mode>_inverted): Add microMIPS support.
14508         (*branch_equality<mode>): Likewise.
14509         (*jump_absolute): Likewise.
14510         (indirect_jump_<mode>): Likewise.
14511         (tablejump_<mode>): Likewise.
14512         (<optab>_internal): Likewise.
14513         (sibcall_internal): Likewise.
14514         (sibcall_value_internal): Likewise.
14515         (prefetch): Use constraint ZD.
14516         * config/mips/mips.opt (minterlink-compressed): New option.
14517         (minterlink-mips16): Now an alias for minterlink-compressed.
14518         (mmicromips): New option.
14519         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14520         (compare_and_swap_12): Likewise.
14521         (sync_add<mode>): Likewise.
14522         (sync_<optab>_12): Likewise.
14523         (sync_old_<optab>_12): Likewise.
14524         (sync_new_<optab>_12): Likewise.
14525         (sync_nand_12): Likewise.
14526         (sync_old_nand_12): Likewise.
14527         (sync_new_nand_12): Likewise.
14528         (sync_sub<mode>): Likewise.
14529         (sync_old_add<mode>): Likewise.
14530         (sync_old_sub<mode>): Likewise.
14531         (sync_new_add<mode>): Likewise.
14532         (sync_new_sub<mode>): Likewise.
14533         (sync_<optab><mode>): Likewise.
14534         (sync_old_<optab><mode>): Likewise.
14535         (sync_new_<optab><mode>): Likewise.
14536         (sync_nand<mode>): Likewise.
14537         (sync_old_nand<mode>): Likewise.
14538         (sync_new_nand<mode>): Likewise.
14539         (sync_lock_test_and_set<mode>): Likewise.
14540         (test_and_set_12): Likewise.
14541         (atomic_compare_and_swap<mode>): Likewise.
14542         (atomic_exchange<mode>_llsc): Likewise.
14543         (atomic_fetch_add<mode>_llsc): Likewise.
14544         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14545         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14546         (umips_save_restore_pattern_p): Likewise.
14547         (umips_load_store_pair_p): Likewise.
14548         (umips_output_load_store_pair): Likewise.
14549         (umips_movep_target_p): Likewise.
14550         (umips_12bit_offset_address_p): Likewise.
14551         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14552         (mips_base_mips16): Rename this...
14553         (mips_base_compression_flags): ...to this. Update all uses.
14554         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14555         (mips_mips16_decl_p): Delete.
14556         (mips_nomips16_decl_p): Delete.
14557         (mips_get_compress_on_flags): New function.
14558         (mips_get_compress_off_flags): New function.
14559         (mips_get_compress_mode): New function.
14560         (mips_get_compress_on_name): New function.
14561         (mips_get_compress_off_name): New function.
14562         (mips_insert_attributes): Support multiple compression types.
14563         (mips_merge_decl_attributes): Likewise.
14564         (umips_12bit_offset_address_p): New function.
14565         (mips_start_function_definition): Emit .set micromips directive.
14566         (mips_call_may_need_jalx_p): New function.
14567         (mips_function_ok_for_sibcall): Add microMIPS support.
14568         (mips_print_operand_punctuation): Support short delay slots and
14569         compact jumps.
14570         (umips_swm_mask, umips_swm_encoding): New.
14571         (umips_build_save_restore): New function.
14572         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14573         (was_mips16_p): Remove.
14574         (old_compression_mode): New.
14575         (mips_set_compression_mode): New function.
14576         (mips_set_current_function): Add microMIPS support.
14577         (mips_option_override): Likewise.
14578         (umips_save_restore_pattern_p): New function.
14579         (umips_output_save_restore): New function.
14580         (umips_load_store_pair_p_1): New function.
14581         (umips_load_store_pair_p): New function.
14582         (umips_output_load_store_pair_1): New function.
14583         (umips_output_load_store_pair): New function.
14584         (umips_movep_target_p) New function.
14585         (mips_prepare_pch_save): Add microMIPS support.
14586         * config/mips/mips.h (TARGET_COMPRESSION): New.
14587         (TARGET_CPU_CPP_BUILTINS): Update macro
14588         to use new compression flags and to support microMIPS.
14589         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14590         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14591         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14592         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14593         (ASM_SPEC): Support mmicromips and mno-micromips.
14594         (M16STORE_REG_P): New macro.
14595         (MIPS_CALL): Support TARGET_MICROMIPS.
14596         (MICROMIPS_J): New macro.
14597         (mips_base_mips16): Rename this...
14598         (mips_base_compression_flags): ...to this.
14599         (UMIPS_12BIT_OFFSET_P): New macro.
14600         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14601         (MULTILIB_DIRNAMES): Likewise.
14602 2013-03-20  Richard Biener  <rguenther@suse.de>
14604         PR tree-optimization/56661
14605         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14606         the result does not have to be distinct.
14608 2013-03-20  Richard Biener  <rguenther@suse.de>
14610         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14611         remap_gimple_op_r.
14613 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14614             Steven Bosscher <steven@gcc.gnu.org>
14616         PR rtl-optimization/56605
14617         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14619 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14621         PR bootstrap/56656
14622         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14623         that require movd instead of movq.
14625 2013-03-20  Richard Biener  <rguenther@suse.de>
14627         * tree-ssa-structalias.c (struct variable_info): Add pointer
14628         to the first field of an aggregate with sub-vars.  Make
14629         this and the pointer to the next subfield its ID.
14630         (vi_next): New function.
14631         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14632         storedanything_id, integer_id): Increment by one.
14633         (new_var_info, get_call_vi, lookup_call_clobber_vi,
14634         get_call_clobber_vi): Adjust.
14635         (solution_set_expand): Simplify and speedup.
14636         (solution_set_add): Inline into ...
14637         (set_union_with_increment): ... this.  Adjust accordingly.
14638         (do_sd_constraint): Likewise.
14639         (do_ds_constraint): Likewise.
14640         (do_complex_constraint): Simplify.
14641         (build_pred_graph): Adjust.
14642         (solve_graph): Likewise.  Simplify and speedup.
14643         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14644         get_constraint_for_component_ref, get_constraint_for_1,
14645         first_vi_for_offset, first_or_preceding_vi_for_offset,
14646         create_function_info_for, create_variable_info_for_1,
14647         create_variable_info_for, intra_create_variable_infos): Adjust.
14648         (init_base_vars): Push NULL for ID zero.
14649         (compute_points_to_sets): Adjust.
14651 2013-03-20  Richard Biener  <rguenther@suse.de>
14653         * cfgloop.c (verify_loop_structure): Streamline and avoid
14654         ICEing on corrupt loop tree.
14655         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14656         loop tree.
14658 2013-03-20  Richard Biener  <rguenther@suse.de>
14660         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14661         check whether an SSA update is needed.
14663 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
14665         * config/mips/constraints.md (T): Rename to...
14666         (Yf): ...this.
14667         (U): Rename to...
14668         (Yd): ...this.
14669         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14670         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14672 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14674         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14675         (*subsi3_carryin_uxtw): Likewise.
14677 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14679         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14680         (*rorsi3_insn_uxtw): Likewise.
14682 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14684         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14685         (*extrsi5_insn_uxtw): Likewise.
14687 2013-03-19  Richard Biener  <rguenther@suse.de>
14689         PR tree-optimization/56273
14690         * passes.c (init_optimization_passes): Move second VRP after DOM.
14692 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
14694         * config/i386/i386.md (*movti_internal): Merge from
14695         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
14696         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
14697         nox64 isa attributes.
14699 2013-03-18  Richard Biener  <rguenther@suse.de>
14701         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14702         (unite): Likewise.
14703         (merge_node_constraints): Likewise.
14704         (build_succ_graph): Likewise.
14705         (valid_graph_edge): Inline into single caller.
14706         (unify_nodes): Likewise.  Use bitmap_set_bit return value
14707         and cache varinfo.
14708         (scc_visit): Fix formatting and variable use.
14709         (do_sd_constraint): Use gcc_checking_assert.
14710         (do_ds_constraint): Likewise.
14711         (do_complex_constraint): Likewise.
14712         (condense_visit): Likewise.  Cleanup.
14713         (dump_pred_graph): New function.
14714         (perform_var_substitution): Dump the pred-graph before
14715         variable substitution.
14716         (find_equivalent_node): Use gcc_checking_assert.
14717         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
14719 2013-03-18  Richard Biener  <rguenther@suse.de>
14721         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
14722         Remove cond_expr_stmt_list argument and do not gimplify the
14723         built expression.
14724         (vect_loop_versioning): Adjust.
14725         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
14726         Cleanup to use less temporaries.
14727         (vect_create_data_ref_ptr): Cleanup.
14729 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14731         PR tree-optimization/56635
14732         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
14733         require types_compatible_p types.
14735 2013-03-18  Nick Clifton  <nickc@redhat.com>
14737         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
14738         spurious backslash.
14740         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
14741         Add missing line to comment describing function.
14743 2013-03-18  Richard Biener  <rguenther@suse.de>
14745         PR tree-optimization/56210
14746         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14747         Handle string / character search functions.
14748         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
14750 2013-03-18  Richard Biener  <rguenther@suse.de>
14752         PR middle-end/56483
14753         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
14754         and implement properly.
14755         * gimple.h (gimple_cond_single_var_p): Remove.
14757 2013-03-18  Richard Biener  <rguenther@suse.de>
14759         * tree-data-ref.h (find_data_references_in_loop): Declare.
14760         * tree-data-ref.c (get_references_in_stmt): Use a stack
14761         vector pre-allocated in the callers.
14762         (find_data_references_in_stmt): Adjust.
14763         (graphite_find_data_references_in_stmt): Likewise.
14764         (create_rdg_vertices): Likewise.
14765         (find_data_references_in_loop): Export.
14766         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14767         Compute dependences here...
14768         (vect_analyze_data_refs): ...not here.  When we encounter
14769         a non-vectorizable data reference in basic-block vectorization
14770         truncate the data reference vector.  Do not bother to
14771         fixup data-dependence information for gather loads.
14772         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
14773         of data references, as reported.
14775 2013-03-18  Richard Biener  <rguenther@suse.de>
14777         PR tree-optimization/3713
14778         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
14779         has_constants and expr.
14780         (stmt_has_constants): Properly valueize SSA names when deciding
14781         whether the stmt has constants.
14783 2013-03-18  Richard Biener  <rguenther@suse.de>
14785         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
14786         whole function when there is nothing to do.
14787         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
14788         * tree-vectorizer.c (vectorize_loops): Update virtual and
14789         loop-closed SSA once.
14790         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
14792 2013-03-18  Richard Biener  <rguenther@suse.de>
14794         PR middle-end/56113
14795         * domwalk.c (bb_postorder): New global static.
14796         (cmp_bb_postorder): New function.
14797         (walk_dominator_tree): Replace scheme imposing an order for
14798         visiting dominator sons by one sorting them at the time they
14799         are pushed on the stack.
14801 2013-03-18  Richard Biener  <rguenther@suse.de>
14803         PR tree-optimization/39326
14804         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14805         (struct mem_ref): Replace mem member with ao_ref typed member.
14806         (MEM_ANALYZABLE): Adjust.
14807         (memref_eq): Likewise.
14808         (mem_ref_alloc): Likewise.
14809         (gather_mem_refs_stmt): Likewise.
14810         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14811         (execute_sm_if_changed_flag_set): Adjust.
14812         (execute_sm): Likewise.
14813         (ref_always_accessed_p): Likewise.
14814         (refs_independent_p): Likewise.
14815         (can_sm_ref_p): Likewise.
14817 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14819         PR c/56566
14820         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14821         return 1 even for !unsignedp.
14823 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14825         * config/i386/i386.md (isa): Add x64 and nox64.
14826         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14827         (*pushtf): Enable *roF alternative for x64 isa only.
14828         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
14829         mode attribute of integer alternatives to DImode for TARGET_64BIT.
14830         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
14831         (*movtf_internal): Merge from *movtf_internal_rex64 and
14832         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
14833         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
14834         nox64 isa attributes.
14835         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
14836         nox64 isa attributes.
14837         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14839 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14841         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14843 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
14845         PR target/56640
14846         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14847         class names.  Remove trailing comma after "ALL_REGS".
14849 2013-03-16  Jan Hubicka  <jh@suse.cz>
14851         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14852         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14853         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14854         of cgraph_get_create_node.
14855         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14857 2013-03-16  Jason Merrill  <jason@redhat.com>
14859         PR debug/49090
14860         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14861         with DW_AT_default_value.
14863 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
14865         * BASE-VER: Set to 4.9.0.
14867 2013-03-14  Andi Kleen  <ak@linux.intel.com>
14869         PR target/56619
14870         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14871         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14872         Document _x* TSX intrinsics.
14874 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
14875             David Holsgrove  <david.holsgrove@xilinx.com>
14877         * configure.ac: Add MicroBlaze TLS support detection.
14878         * configure: Regenerate.
14879         * config/microblaze/microblaze-protos.h
14880         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14881         symbol_mentioned_p, label_mentioned_p): Add prototypes.
14882         * config/microblaze/microblaze.c (microblaze_address_type): Add
14883         ADDRESS_TLS and tls_reloc address types.
14884         (microblaze_address_info): Add tls_reloc.
14885         (TARGET_HAVE_TLS): Define.
14886         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14887          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14888          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14889         load_tls_operand,  microblaze_call_tls_get_addr,
14890         microblaze_legitimize_tls_address): New functions.
14891         (microblaze_classify_unspec): Handle UNSPEC_TLS.
14892         (get_base_reg): Use microblaze_tls_symbol_p.
14893         (microblaze_classify_address): Handle TLS.
14894         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14895         label_mentioned_p and microblaze_tls_referenced_p.
14896         (microblaze_legitimize_address): Handle TLS.
14897         (microblaze_address_insns): Handle ADDRESS_TLS.
14898         (pic_address_needs_scratch): Handle TLS.
14899         (print_operand_address): Handle TLS.
14900         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14901         (microblaze_expand_move): Handle TLS.
14902         (microblaze_legitimate_constant_p): Check
14903         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14904         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14905         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14906         (PIC_OFFSET_TABLE_REGNUM): Set.
14907         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14908         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14909         (addsi3, movsi_internal2, movdf_internal): Update constraints
14910         * config/microblaze/predicates.md (arith_plus_operand): Define
14911         (move_operand): Redefine as move_src_operand,
14912         check microblaze_tls_referenced_p.
14914 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14916         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14917         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14919 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14921         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14922         CC mode for AND.
14924 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14926         PR tree-optimization/53265
14927         * common.opt (Waggressive-loop-optimizations): New option.
14928         * tree-ssa-loop-niter.c: Include tree-pass.h.
14929         (do_warn_aggressive_loop_optimizations): New function.
14930         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14931         if number_of_latch_executions returned constant.
14932         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14933         early.  If number_of_latch_executions returned constant, set
14934         nb_iterations_upper_bound back to it.
14935         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14936         field.
14937         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14938         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14940         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14941         (MULTILIB_OSDIRNAMES): Set.
14942         * genmultilib: If defaultosdirname doesn't start with :: , set
14943         defaultosdirname2 instead, clear it and emit two . multilib_raw
14944         entries instead of just one.
14946 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
14948         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14949         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14950         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14951         (SUBTARGET_OVERRIDE_OPTIONS): New.
14953 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14955         PR target/49880
14956         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14957         (musermode): Convert to Var(TARGET_USERMODE).
14958         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14959         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14960         * config/sh/sh.c (sh_option_override): Use
14961         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14962         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14963         condition.
14964         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14965         TARGET_SH4.
14966         (udivsi3_i4_single, divsi3_i4_single): Use
14967         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14969 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14971         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14972         default setting.
14974 2013-03-13  Richard Biener  <rguenther@suse.de>
14976         PR tree-optimization/56608
14977         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14978         calls when vectorizing basic-blocks.
14980 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14982         PR plugins/45078
14983         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14984         tm_file.
14986 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14988         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14990 2013-03-11  Jan Hubicka  <jh@suse.cz>
14992         PR lto/56557
14993         * lto-streamer-out.c (output_symbol_p): Skip references from
14994         constructors of external variables.
14996 2013-03-11  Jan Hubicka  <jh@suse.cz>
14998         PR middle-end/56571
14999         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15000         from pseudos.
15001         * emit-rtl.c (verify_rtx_sharing): Likewise.
15002         (copy_insn_1): Likewise.
15003         * rtl.c (copy_rtx): Likewise.
15005 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
15007         PR target/56591
15008         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15009         output_operand_lossage message.
15011 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
15013         PR target/56470
15014         * arm.c (shift_op): Validate RTL pattern on the fly.
15015         (arm_print_operand, case 'S'): Don't use shift_operator to validate
15016         the RTL.
15018 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15020         PR target/56347
15021         * config/pa/pa.md (call_value): Check for calls to powf and direct to
15022         new call patterns that clobber %fr12.
15023         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15024         split and postreload patterns.
15025         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15026         registers %fr12 and %fr12R as call used.
15028 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
15030         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15031         (canon_address, record_store, replace_read, check_mem_read_rtx,
15032         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15033         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15034         rest_of_handle_dse): Likewise.
15036 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
15038         PR middle-end/56524
15039         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15040         Add base_optabs.
15041         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15042         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15043         (save_optabs_if_changed): Replace with...
15044         (init_tree_optimization_optabs): ...this.
15045         * optabs.c (save_optabs_if_changed): Rename to...
15046         (init_tree_optimization_optabs): ...this.  Take the optimization node
15047         as argument.  Do nothing if the base optabs are already correct.
15048         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15049         to recompute optabs.
15050         * function.h (function): Remove optabs field.
15051         * function.c (invoke_set_current_function_hook): Call
15052         init_tree_optimization_optabs.  Use the result to initialize
15053         this_fn_optabs.
15055 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
15057         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15058         if GTMA_HAS_NO_INSTRUMENTATION.
15059         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15060         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15061         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15062         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15063         GTMA_HAS_NO_INSTRUMENTATION.
15065 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15067         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15068         libasan_preinit.o.
15070 2013-03-08  Marek Polacek  <polacek@redhat.com>
15071             Jakub Jelinek  <jakub@redhat.com>
15073         PR tree-optimization/56478
15074         * predict.c (is_comparison_with_loop_invariant_p): Change the
15075         type of loop_step to tree.
15076         (predict_loops): Adjust.
15077         (predict_iv_comparison): Perform the computations on double_ints.
15079 2013-03-08  Richard Biener  <rguenther@suse.de>
15081         PR tree-optimization/56570
15082         * tree-cfg.c (verify_expr_location_1): Verify locations for
15083         DECL_DEBUG_EXPR.
15084         * tree-sra.c (create_access_replacement): Strip locations
15085         from DECL_DEBUG_EXPRs.
15087 2013-03-08  Richard Biener  <rguenther@suse.de>
15089         * tree-inline.c (expand_call_inline): Do not associate
15090         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15091         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15093 2013-03-08  Richard Biener  <rguenther@suse.de>
15095         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15096         or block changes with -Og.  Fix for location / block encoding
15097         changes and PHI arguments with locations.
15099 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15101         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15102         for all counters.
15103         (struct output_info): Likewise.
15104         (register_overhead): Remove bad gcc_assert.
15105         (bitmap_find_bit): If there is only a single bitmap element, do not
15106         count a miss as a search.
15107         (print_statistics): Update for counter type changes.
15108         (dump_bitmap_statistics): Likewise.  Print headers such that they
15109         are properly lined up with the printed counters.
15111 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
15113         PR tree-optimization/56559
15114         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15115         check that it has only a single use.
15117 2013-03-07  Richard Biener  <rguenther@suse.de>
15119         * doc/invoke.texi (fwhole-program): Discourage use in combination
15120         with -flto.
15122 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
15124         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15126         PR tree-optimization/56539
15127         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15128         instead of GSI_CONTINUE_LINKING as last argument to
15129         force_gimple_operand_gsi.  Adjust function comment.
15131         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15132         aarch64-cores.def.
15134         PR middle-end/56548
15135         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15136         promoted mode, convert the result back to the original mode.
15138 2013-03-06  Richard Biener  <rguenther@suse.de>
15140         PR middle-end/56294
15141         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15142         (insert_updated_phi_nodes_compare_uids): New function.
15143         (update_ssa): Sort symbols_to_rename after UID before
15144         traversing it to insert PHI nodes.
15146 2013-03-06  Richard Biener  <rguenther@suse.de>
15148         PR middle-end/50494
15149         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15150         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15152         Revert
15153         2013-02-13  Richard Biener  <rguenther@suse.de>
15155         PR lto/50494
15156         * varasm.c (output_constant_def_1): Get the decl representing
15157         the constant as argument.
15158         (output_constant_def): Wrap output_constant_def_1.
15159         (make_decl_rtl): Use output_constant_def_1 with the decl
15160         representing the constant.
15161         (build_constant_desc): Optionally re-use a decl already
15162         representing the constant.
15163         (tree_output_constant_def): Adjust.
15165 2013-03-06  Joey Ye  <joey.ye@arm.com>
15167         PR lto/50293
15168         * gcc.c (convert_white_space): New function.
15169         (main): Handles white space in function name.
15171 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
15173         PR target/56529
15174         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15175         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
15176         to SH_DIV_CALL_TABLE for TARGET_SH2.
15177         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15178         list.
15179         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15180         call-table options.
15182 2013-03-05  Sterling Augustine  <saugustine@google.com>
15183             Cary Coutant  <ccoutant@google.com>
15185         PR debug/55364
15186         * dwarf2out.c (resolve_addr): Don't call
15187         remove_loc_list_addr_table_entries a second time for the same
15188         expression.
15190 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15192         PR debug/56510
15193         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15194         (avoid_complex_debug_insns): New function.
15195         (expand_debug_locations): Call it.
15197         PR rtl-optimization/56484
15198         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15199         lifetimes of hard registers on small register class machines.
15201 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15203         * config/microblaze/microblaze-protos.h: Rename
15204         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15205         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15206         fast_interrupt.
15207         (microblaze_fast_interrupt_function_p): New function.
15208         (microblaze_is_interrupt_handler): Rename to
15209         microblaze_is_interrupt_variant and add fast_interrupt check.
15210         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15211         (save_restore_insns): Likewise.
15212         (compute_frame_size): Likewise.
15213         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15214         (microblaze_globalize_label): Likewise.
15215         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15216         * config/microblaze/microblaze.md: Use wrapper
15217         microblaze_is_interrupt_variant.
15219 2013-03-05  Kai Tietz  <ktietz@redhat.com>
15221         * sdbout.c (sdbout_one_type): Switch to current function's section
15222         supporting cold/hot.
15224 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15226         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15227         -mxl-reorder.
15229 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15231         PR middle-end/56461
15232         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15233         if VALGRIND_GET_VBITS is defined, temporarily make object
15234         memory all defined, and restore previous valgrind addressability
15235         and definability afterwards.  Free this_object at the end.
15237         PR middle-end/56461
15238         * lra.c (lra): Call lra_clear_live_ranges if live_p,
15239         right before calling lra_create_live_ranges, also call it
15240         when clearing live_p.  Only call lra_clear_live_ranges
15241         at the end if live_p.
15243         PR middle-end/56461
15244         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15246 2013-03-05  Richard Biener  <rguenther@suse.de>
15248         PR tree-optimization/56521
15249         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15250         value-id.
15252 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
15254         PR c++/55135
15255         * except.h (remove_unreachable_eh_regions): New prototype.
15256         * except.c (remove_eh_handler_splicer): New function, split out
15257         of remove_eh_handler.
15258         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
15259         warning about running it on many EH regions one at a time.
15260         (remove_unreachable_eh_regions_worker): New function, walk the
15261         EH tree in depth-first order and remove non-marked regions.
15262         (remove_unreachable_eh_regions): New function.
15263         * tree-eh.c (mark_reachable_handlers): New function, split out
15264         from remove_unreachable_handlers.
15265         (remove_unreachable_handlers): Use mark_reachable_handlers and
15266         remove_unreachable_eh_regions.
15267         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15268         and remove_unreachable_eh_regions.
15270 2013-03-05  Richard Biener  <rguenther@suse.de>
15272         PR middle-end/56525
15273         * loop-init.c (fix_loop_structure): Remove loops in two stages,
15274         not freeing them until the end.
15276 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15278         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15280 2013-03-05  Richard Biener  <rguenther@suse.de>
15282         PR tree-optimization/56270
15283         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15284         of loads after scheduling an SLP instance.
15286 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15288         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15289         tic6x.exp.
15290         (check_gcc_parallelize): Run guality.exp as a separate job from
15291         vect.exp with unsorted.exp and $(dg_target_exps) separately from
15292         struct-layout-1.exp with stackalign.exp.
15294         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15296         PR middle-end/56461
15297         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15298         load_index sbitmap even if some bit in it isn't set.
15300         PR middle-end/56461
15301         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
15302         (discover_iteration_bound_by_body_walk): Change queues to
15303         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
15304         spelling in comment.  Call safe_push on queues[bound_index] directly.
15305         Release queues[queue_index] in every iteration unconditionally.
15306         Release bounds vector.
15308         PR middle-end/56461
15309         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
15310         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
15311         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
15312         inner_phis vector.
15314 2013-03-05  Richard Biener  <rguenther@suse.de>
15316         PR lto/56515
15317         * tree-inline.c (remap_blocks_to_null): New function.
15318         (expand_call_inline): When expanding a call stmt without
15319         an associated block inline remap all callee blocks to NULL.
15321 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15323         PR rtl-optimization/56494
15324         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
15325         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
15326         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
15328         PR middle-end/56461
15329         * sel-sched-ir.c (free_sched_pools): Release
15330         succs_info_pool.stack[succs_info_pool.max_top] vectors too
15331         if succs_info_pool.max_top isn't -1.
15333         PR bootstrap/56509
15334         * opts.c (opts_obstack, opts_concat): Moved to...
15335         * opts-common.c (opts_obstack, opts_concat): ... here.
15337 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15339         PR middle-end/56461
15340         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
15342 2013-03-04  Martin Jambor  <mjambor@suse.cz>
15344         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
15345         all appropriate places.
15347 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
15349         PR tree-optimization/56424
15350         * ipa-split.c (split_function): Do not set the RSO flag if result is
15351         not by reference and its type is a register type.
15353 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
15355         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
15356         (microblaze_legitimate_pic_operand): Likewise
15357         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
15358         new function microblaze_legitimate_pic_operand
15359         * config/microblaze/microblaze-protos.h
15360         (microblaze_legitimate_pic_operand): Declare.
15362 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15364         * config/microblaze/predicates.md (call_insn_simple_operand):
15365         New predicate for supported rtx code types.
15366         * config/microblaze/microblaze.md (call_internal1): Use
15367         call_insn_simple_operand predicate.
15369 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15371         PR middle-end/56461
15372         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
15373         partitions.ordered_remove.
15375         PR middle-end/56461
15376         * tree-vect-stmts.c (vectorizable_conversion): Don't call
15377         vec_oprnds0.create (1) for modifier == NONE.
15379         PR middle-end/56461
15380         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
15381         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
15382         vec_oprnds1 right before pushing anything to it for
15383         scalar_shift_arg.
15385         PR middle-end/56461
15386         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
15387         set nbbs to 0 instead of having separate code path.
15388         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
15389         instead of false as last argument if returning NULL.
15391 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
15393         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
15394         the attribute is now called "target" instead of "option".
15395         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
15396         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
15397         attribute/pragma name for TARGET_OPTION_VALID_P and
15398         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
15399         * doc/tm.texi: Regenerated.
15401 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
15403         * config/microblaze/microblaze.c:
15404         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
15405         * config/microblaze/microblaze.h: Add -mxl-reorder to
15406         DRIVER_SELF_SPECS.
15407         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
15408         instructions emitted if TARGET_REORDER.
15409         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
15410         or 0 for -m/-mno case, but initialises as 2 to detect default use case
15411         separately.
15413 2013-03-01  Xinliang David Li  <davidxl@google.com>
15415         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
15416         walk length.
15418 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
15420         PR middle-end/56461
15421         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
15422         vector even when returning true.  Fix up function comment formatting.
15424         PR middle-end/56461
15425         * ira-build.c (ira_loop_nodes_count): New variable.
15426         (create_loop_tree_nodes): Initialize it.
15427         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
15429         PR middle-end/56461
15430         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
15431         method on dr_chain and result_chain.
15432         * tree-vect-stmts.c (vectorizable_store): Only call
15433         result_chain.create if j == 0.
15435         PR middle-end/56461
15436         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
15437         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
15438         before overwriting it.
15440 2013-03-01  Tobias Burnus  <burnus@net-b.de>
15442         * doc/extended.texi (C Extensions): Change order in @menu
15443         to match @node.
15444         (Other MIPS Built-in Functions): Move last MIPS entry before
15445         "picoChip Built-in Functions".
15446         (SH Built-in Functions): Move after RX Built-in Functions.
15447         * doc/gcc.texi (Introduction): Change order in @menu
15448         to match @node.
15449         * doc/md.texi (Constraints): Ditto.
15450         * gty.texi (Type Information): Ditto.
15451         (User-provided marking routines for template types): Make
15452         subsection.
15453         * doc/invoke.texi (AArch64 Options): Move before
15454         "Adapteva Epiphany Options".
15456 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
15457             Jakub Jelinek  <jakub@redhat.com>
15459         PR sanitizer/56454
15460         * asan.c (gate_asan): Lookup no_sanitize_address instead of
15461         no_address_safety_analysis attribute.
15462         * doc/extend.texi (no_address_safety_attribute): Rename to
15463         no_sanitize_address attribute, mention no_address_safety_analysis
15464         attribute as deprecated alias.
15466 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
15468         PR middle-end/56461
15469         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
15470         type to vec<vec<tree> > *.
15471         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
15472         to be vec<tree> instead of vec<tree> *, set vec_defs
15473         to vNULL and call vec_defs.create (number_of_vects), adjust other
15474         uses of vec_defs.
15475         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
15476         vectorizable_condition): Adjust vect_get_slp_defs callers.
15478 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15480         * config/aarch64/aarch64.c
15481         (aarch64_float_const_representable): Remove unused variable.
15483 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15485         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
15487 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15489         * config/aarch64/aarch64-builtins.c
15490         (aarch64_init_simd_builtins): Make static.
15492 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15494         * config/aarch64/aarch64.c
15495         (aarch64_simd_make_constant): Make static.
15497 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15499         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
15500         with no initialization to the RHS of debug statements.
15502 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15504         PR tree-optimization/56294
15505         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
15506         Adjust dumping.
15507         (get_access_replacement): Do not call create_access_replacement.
15508         Assert a replacement exists.
15509         (get_repl_default_def_ssa_name): Create the replacement declaration
15510         itself.
15512 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15514         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
15515         final_end_function.
15517 2013-02-28  Marek Polacek  <polacek@redhat.com>
15519         PR rtl-optimization/56466
15520         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
15521         if we're changing a loop.
15522         (peel_loops_completely): Likewise.
15524 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
15526         PR c++/55813
15527         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15529 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15531         PR target/56445
15532         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
15533         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
15534         INTX_FTYPE_FX, FX_FTYPE_INTX.
15535         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
15537 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15539         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
15540         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
15541         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15542         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15543         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15544         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15545         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15546         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15547         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15548         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15549         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15550         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15551         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15552         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15553         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15554         (avrxmega6): Increase max flash segments from 5 to 6.
15555         * config/avr/t-multilib: Regenerate.
15556         * config/avr/avr-tables.opt: Regenerate.
15557         * doc/avr-mmcu.texi: Regenerate.
15559 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15561         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15562         (avr_device_to_arch): Rename to avr_device_to_ld.
15563         (avr_device_to_as): New prototype.
15564         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15565         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15566         * config/avr/driver-avr.c (avr_device_to_as): New.
15567         (avr_device_to_arch): Rename to avr_device_to_ld.
15569 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15571         PR middle-end/56461
15572         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15573         method on dr_chain and result_chain.
15575         PR middle-end/56461
15576         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15577         pointer_set_destroy on not_executed_last_iteration.
15579         PR middle-end/56461
15580         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15582         PR middle-end/56461
15583         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15584         FOR_EACH_DEFINED_FUNCTION when freeing state.
15586         PR middle-end/56461
15587         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15588         pool_free.
15589         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15590         overwriting it.
15592         PR middle-end/56461
15593         * ipa-cp.c (decide_whether_version_node): Call vec_free on
15594         known_aggs[i].items and release known_aggs vector.
15596         PR middle-end/56461
15597         * ipa-reference.c (propagate): Free node_info even for alias nodes.
15599 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15601         * config/microblaze/microblaze.c (microblaze_emit_compare):
15602         Use xor for EQ/NE comparisions.
15603         * config/microblaze/microblaze.md (cstoresf4): Add constraints
15604         (cbranchsf4): Adjust operator to comparison_operator.
15606 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15608         PR middle-end/56461
15609         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15610         vector.
15611         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15612         vec_safe_push, always update *slot.
15613         (redirect_edge_var_map_clear): Use vec_free.
15614         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15615         (free_var_map_entry): Use vec_free.
15616         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15617         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15619 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
15621         PR middle-end/45472
15622         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15623         when the may_trap_p bit of the exprs being merged differs.
15624         Reorder tests for speculativeness in the logical and operator.
15626 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15628         * incpath.c (add_standard_paths): Use reconcat instead of concat
15629         where appropriate and avoid leaking memory.
15631         * opts.h: Include obstack.h.
15632         (opts_concat): New prototype.
15633         (opts_obstack): New declaration.
15634         * opts.c (opts_concat): New function.
15635         (opts_obstack): New variable.
15636         (init_options_struct): Call gcc_init_obstack on opts_obstack.
15637         (finish_options): Use opts_concat instead of concat
15638         and XOBNEWVEC instead of XNEWVEC.
15639         * opts-common.c (generate_canonical_option, decode_cmdline_option,
15640         generate_option): Likewise.
15641         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15642         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15644         PR target/56455
15645         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15646         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15648 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15650         PR middle-end/56461
15651         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15653 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15655         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15656         (arm_block_move_unaligned_straight): Likewise.
15657         (arm_adjust_block_mem): Likewise.
15659 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15661         PR target/48901
15662         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15663         temp, cond and label.
15664         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15666         PR target/52500
15667         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15668         * config/c6x/c6x.h (dbx_register_map): Update declaration.
15670         PR target/52501
15671         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15672         of prologue/epilogue functions.
15674         PR target/52550
15675         * config/tilegx/tilegx.c (tilegx_expand_prologue):
15676         Remove unused variable cfa_offset.
15677         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15679         PR target/54639
15680         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15681         type promotion to unsigned.
15683         PR target/54640
15684         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15685         for HOST_WIDE_INT of 32 bit / same size as int.
15686         (arm_block_move_unaligned_straight): Likewise.
15687         (arm_adjust_block_mem): Likewise.
15689         PR target/54662
15690         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15691         ALL_CFLAGS.
15693 2013-02-26  Marek Polacek  <polacek@redhat.com>
15695         PR tree-optimization/56426
15696         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15698 2013-02-26  Richard Biener  <rguenther@suse.de>
15700         PR target/56444
15701         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15702         unused variable loops.
15704 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15706         PR tree-optimization/56448
15707         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
15708         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
15709         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
15710         later operands of the references, or even first operand for
15711         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
15713         PR tree-optimization/56443
15714         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
15715         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
15716         to type_for_mode langhook.
15718 2013-02-25  Matt Turner  <mattst88@gmail.com>
15720         * doc/invoke.texi: Document r4700.
15722 2013-02-25  Richard Biener  <rguenther@suse.de>
15724         PR tree-optimization/56175
15725         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
15726         split out from ...
15727         (simplify_bitwise_binary): ... here.  Also guard the conversion
15728         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
15730 2013-02-25  Catherine Moore  <clm@codesourcery.com>
15732         Revert:
15733         2013-02-24  Catherine Moore  <clm@codesourcery.com>
15734             Maciej W. Rozycki  <macro@codesourcery.com>
15735             Tom de Vries  <tom@codesourcery.com>
15736             Nathan Sidwell  <nathan@codesourcery.com>
15737             Iain Sandoe  <iain@codesourcery.com>
15738             Nathan Froyd  <froydnj@codesourcery.com>
15739             Chao-ying Fu  <fu@mips.com>
15741         * doc/extend.texi: (micromips, nomicromips, nocompression):
15742         Document new function attributes.
15743         * doc/invoke.texi (minterlink-compressed, mmicromips,
15744         m14k, m14ke, m14kec): Document new options.
15745         (minterlink-mips16): Update documentation.
15746         * doc/md.texi (ZC, ZD): Document new constraints.
15747         * configure.ac (gcc_cv_as_micromips): Check if linker
15748         supports the .set micromips directive.
15749         * configure: Regenerate.
15750         * config.in: Regenerate.
15751         * config/mips/mips-tables.opt: Regenerate.
15752         * config/mips/micromips.md: New file.
15753         * constraints.md (ZC, AD): New constraints.
15754         * config/mips/predicates.md (movep_src_register): New predicate.
15755         (movep_src_operand): New predicate.
15756         (non_volatile_mem_operand): New predicate.
15757         * config/mips/mips.md (multimem): New type.
15758         (length): Differentiate between 17-bit and 18-bit branch offsets.
15759         (MOVEP1, MOVEP2): New mode iterator.
15760         (mov_<load>l): Use ZC constraint.
15761         (mov_<load>r): Likewise.
15762         (mov_<store>l): Likewise.
15763         (mov_<store>r): Likewise.
15764         (*branch_equality<mode>_inverted): Add microMIPS support.
15765         (*branch_equality<mode>): Likewise.
15766         (*jump_absolute): Likewise.
15767         (indirect_jump_<mode>): Likewise.
15768         (tablejump_<mode>): Likewise.
15769         (<optab>_internal): Likewise.
15770         (sibcall_internal): Likewise.
15771         (sibcall_value_internal): Likewise.
15772         (prefetch): Use constraint ZD.
15773         * config/mips/mips.opt (minterlink-compressed): New option.
15774         (minterlink-mips16): Now an alias for minterlink-compressed.
15775         (mmicromips): New option.
15776         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15777         (compare_and_swap_12): Likewise.
15778         (sync_add<mode>): Likewise.
15779         (sync_<optab>_12): Likewise.
15780         (sync_old_<optab>_12): Likewise.
15781         (sync_new_<optab>_12): Likewise.
15782         (sync_nand_12): Likewise.
15783         (sync_old_nand_12): Likewise.
15784         (sync_new_nand_12): Likewise.
15785         (sync_sub<mode>): Likewise.
15786         (sync_old_add<mode>): Likewise.
15787         (sync_old_sub<mode>): Likewise.
15788         (sync_new_add<mode>): Likewise.
15789         (sync_new_sub<mode>): Likewise.
15790         (sync_<optab><mode>): Likewise.
15791         (sync_old_<optab><mode>): Likewise.
15792         (sync_new_<optab><mode>): Likewise.
15793         (sync_nand<mode>): Likewise.
15794         (sync_old_nand<mode>): Likewise.
15795         (sync_new_nand<mode>): Likewise.
15796         (sync_lock_test_and_set<mode>): Likewise.
15797         (test_and_set_12): Likewise.
15798         (atomic_compare_and_swap<mode>): Likewise.
15799         (atomic_exchange<mode>_llsc): Likewise.
15800         (atomic_fetch_add<mode>_llsc): Likewise.
15801         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15802         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15803         (umips_save_restore_pattern_p): Likewise.
15804         (umips_load_store_pair_p): Likewise.
15805         (umips_output_load_store_pair): Likewise.
15806         (umips_movep_target_p): Likewise.
15807         (umips_12bit_offset_address_p): Likewise.
15808         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15809         (mips_base_mips16): Rename this...
15810         (mips_base_compression_flags): ...to this. Update all uses.
15811         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15812         (mips_mips16_decl_p): Delete.
15813         (mips_nomips16_decl_p): Delete.
15814         (mips_get_compress_on_flags): New function.
15815         (mips_get_compress_off_flags): New function.
15816         (mips_get_compress_mode): New function.
15817         (mips_get_compress_on_name): New function.
15818         (mips_get_compress_off_name): New function.
15819         (mips_insert_attributes): Support multiple compression types.
15820         (mips_merge_decl_attributes): Likewise.
15821         (umips_12bit_offset_address_p): New function.
15822         (mips_start_function_definition): Emit .set micromips directive.
15823         (mips_call_may_need_jalx_p): New function.
15824         (mips_function_ok_for_sibcall): Add microMIPS support.
15825         (mips_print_operand_punctuation): Support short delay slots and
15826         compact jumps.
15827         (umips_swm_mask, umips_swm_encoding): New.
15828         (umips_build_save_restore): New function.
15829         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15830         (was_mips16_p): Remove.
15831         (old_compression_mode): New.
15832         (mips_set_compression_mode): New function.
15833         (mips_set_current_function): Add microMIPS support.
15834         (mips_option_override): Likewise.
15835         (umips_save_restore_pattern_p): New function.
15836         (umips_output_save_restore): New function.
15837         (umips_load_store_pair_p_1): New function.
15838         (umips_load_store_pair_p): New function.
15839         (umips_output_load_store_pair_1): New function.
15840         (umips_output_load_store_pair): New function.
15841         (umips_movep_target_p) New function.
15842         (mips_prepare_pch_save): Add microMIPS support.
15843         * config/mips/mips.h (TARGET_COMPRESSION): New.
15844         (TARGET_CPU_CPP_BUILTINS): Update macro
15845         to use new compression flags and to support microMIPS.
15846         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15847         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15848         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15849         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15850         (ASM_SPEC): Support mmicromips and mno-micromips.
15851         (M16STORE_REG_P): New macro.
15852         (MIPS_CALL): Support TARGET_MICROMIPS.
15853         (MICROMIPS_J): New macro.
15854         (mips_base_mips16): Rename this...
15855         (mips_base_compression_flags): ...to this.
15856         (UMIPS_12BIT_OFFSET_P): New macro.
15857         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15858         (MULTILIB_DIRNAMES): Likewise.
15860 2013-02-25  Tom de Vries  <tom@codesourcery.com>
15862         PR rtl-optimization/56131
15863         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15864         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15865         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15867 2013-02-25  Tobias Burnus  <burnus@net-b.de>
15869         * doc/invoke.texi (-fsanitize=): Move from optimization
15870         to debugging options.
15872 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15874         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15876 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15877             Alexander Monakov  <amonakov@ispras.ru>
15879         PR middle-end/56077
15880         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15881         flush pending lists also on non-jumps.  Adjust comment.
15883 2013-02-24  Catherine Moore  <clm@codesourcery.com>
15884             Maciej W. Rozycki  <macro@codesourcery.com>
15885             Tom de Vries  <tom@codesourcery.com>
15886             Nathan Sidwell  <nathan@codesourcery.com>
15887             Iain Sandoe  <iain@codesourcery.com>
15888             Nathan Froyd  <froydnj@codesourcery.com>
15889             Chao-ying Fu  <fu@mips.com>
15891         * doc/extend.texi: (micromips, nomicromips, nocompression):
15892         Document new function attributes.
15893         * doc/invoke.texi (minterlink-compressed, mmicromips,
15894         m14k, m14ke, m14kec): Document new options.
15895         (minterlink-mips16): Update documentation.
15896         * doc/md.texi (ZC, ZD): Document new constraints.
15897         * configure.ac (gcc_cv_as_micromips): Check if linker
15898         supports the .set micromips directive.
15899         * configure: Regenerate.
15900         * config.in: Regenerate.
15901         * config/mips/mips-tables.opt: Regenerate.
15902         * config/mips/micromips.md: New file.
15903         * constraints.md (ZC, AD): New constraints.
15904         * config/mips/predicates.md (movep_src_register): New predicate.
15905         (movep_src_operand): New predicate.
15906         (non_volatile_mem_operand): New predicate.
15907         * config/mips/mips.md (multimem): New type.
15908         (length): Differentiate between 17-bit and 18-bit branch offsets.
15909         (MOVEP1, MOVEP2): New mode iterator.
15910         (mov_<load>l): Use ZC constraint.
15911         (mov_<load>r): Likewise.
15912         (mov_<store>l): Likewise.
15913         (mov_<store>r): Likewise.
15914         (*branch_equality<mode>_inverted): Add microMIPS support.
15915         (*branch_equality<mode>): Likewise.
15916         (*jump_absolute): Likewise.
15917         (indirect_jump_<mode>): Likewise.
15918         (tablejump_<mode>): Likewise.
15919         (<optab>_internal): Likewise.
15920         (sibcall_internal): Likewise.
15921         (sibcall_value_internal): Likewise.
15922         (prefetch): Use constraint ZD.
15923         * config/mips/mips.opt (minterlink-compressed): New option.
15924         (minterlink-mips16): Now an alias for minterlink-compressed.
15925         (mmicromips): New option.
15926         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15927         (compare_and_swap_12): Likewise.
15928         (sync_add<mode>): Likewise.
15929         (sync_<optab>_12): Likewise.
15930         (sync_old_<optab>_12): Likewise.
15931         (sync_new_<optab>_12): Likewise.
15932         (sync_nand_12): Likewise.
15933         (sync_old_nand_12): Likewise.
15934         (sync_new_nand_12): Likewise.
15935         (sync_sub<mode>): Likewise.
15936         (sync_old_add<mode>): Likewise.
15937         (sync_old_sub<mode>): Likewise.
15938         (sync_new_add<mode>): Likewise.
15939         (sync_new_sub<mode>): Likewise.
15940         (sync_<optab><mode>): Likewise.
15941         (sync_old_<optab><mode>): Likewise.
15942         (sync_new_<optab><mode>): Likewise.
15943         (sync_nand<mode>): Likewise.
15944         (sync_old_nand<mode>): Likewise.
15945         (sync_new_nand<mode>): Likewise.
15946         (sync_lock_test_and_set<mode>): Likewise.
15947         (test_and_set_12): Likewise.
15948         (atomic_compare_and_swap<mode>): Likewise.
15949         (atomic_exchange<mode>_llsc): Likewise.
15950         (atomic_fetch_add<mode>_llsc): Likewise.
15951         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15952         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15953         (umips_save_restore_pattern_p): Likewise.
15954         (umips_load_store_pair_p): Likewise.
15955         (umips_output_load_store_pair): Likewise.
15956         (umips_movep_target_p): Likewise.
15957         (umips_12bit_offset_address_p): Likewise.
15958         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15959         (mips_base_mips16): Rename this...
15960         (mips_base_compression_flags): ...to this. Update all uses.
15961         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15962         (mips_mips16_decl_p): Delete.
15963         (mips_nomips16_decl_p): Delete.
15964         (mips_get_compress_on_flags): New function.
15965         (mips_get_compress_off_flags): New function.
15966         (mips_get_compress_mode): New function.
15967         (mips_get_compress_on_name): New function.
15968         (mips_get_compress_off_name): New function.
15969         (mips_insert_attributes): Support multiple compression types.
15970         (mips_merge_decl_attributes): Likewise.
15971         (umips_12bit_offset_address_p): New function.
15972         (mips_start_function_definition): Emit .set micromips directive.
15973         (mips_call_may_need_jalx_p): New function.
15974         (mips_function_ok_for_sibcall): Add microMIPS support.
15975         (mips_print_operand_punctuation): Support short delay slots and
15976         compact jumps.
15977         (umips_swm_mask, umips_swm_encoding): New.
15978         (umips_build_save_restore): New function.
15979         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15980         (was_mips16_p): Remove.
15981         (old_compression_mode): New.
15982         (mips_set_compression_mode): New function.
15983         (mips_set_current_function): Add microMIPS support.
15984         (mips_option_override): Likewise.
15985         (umips_save_restore_pattern_p): New function.
15986         (umips_output_save_restore): New function.
15987         (umips_load_store_pair_p_1): New function.
15988         (umips_load_store_pair_p): New function.
15989         (umips_output_load_store_pair_1): New function.
15990         (umips_output_load_store_pair): New function.
15991         (umips_movep_target_p) New function.
15992         (mips_prepare_pch_save): Add microMIPS support.
15993         * config/mips/mips.h (TARGET_COMPRESSION): New.
15994         (TARGET_CPU_CPP_BUILTINS): Update macro
15995         to use new compression flags and to support microMIPS.
15996         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15997         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15998         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15999         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16000         (ASM_SPEC): Support mmicromips and mno-micromips.
16001         (M16STORE_REG_P): New macro.
16002         (MIPS_CALL): Support TARGET_MICROMIPS.
16003         (MICROMIPS_J): New macro.
16004         (mips_base_mips16): Rename this...
16005         (mips_base_compression_flags): ...to this.
16006         (UMIPS_12BIT_OFFSET_P): New macro.
16007         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16008         (MULTILIB_DIRNAMES): Likewise.
16010 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
16012         PR target/52555
16013         * target-globals.c (save_target_globals): For init_reg_sets and
16014         target_reinit remporarily set this_fn_optabs to this_target_optabs.
16016 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
16018         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16019         * config/aarch64/t-aarch64
16020         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16022 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
16024         PR inline-asm/56148
16025         * lra-constraints.c (process_alt_operands): Reload operand
16026         conflicting with earlier clobber only if no more other conflicting
16027         operands.
16029 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
16031         PR sanitizer/56393
16032         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16033         if not linking a shared library.
16035 2013-02-22  Seth LaForge  <sethml@google.com>
16037         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16039 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
16041         * config/arm/arm.md (split for extendsidi): Update condition.
16042         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16043         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16044         (qhs_zextenddi_cstr): Likewise.
16046 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16048         PR middle-end/56420
16049         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16050         avoid signed wrapping.
16051         (expand_mult): Handle properly multiplication by
16052         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
16053         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
16054         in the compiler if coeff is HOST_WIDE_INT_MIN.
16055         (expand_divmod): Don't make ext_op1 static, change it's type to
16056         uhwi.  Avoid undefined behavior in -INTVAL (op1).
16058         PR rtl-optimization/50339
16059         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16060         field.
16061         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16062         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16063         into splitting_ashiftrt field.
16064         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16065         ASHIFTRT.
16066         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16067         choices.
16069 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16071         PR middle-end/56108
16072         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16073         are sure to go irrevocable.
16075 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16077         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16078         scalars are valid operands.
16080 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16082         PR tree-optimization/56310
16083         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16084         only matching indices and non-negative final offsets.
16085         (intersect_aggregates_with_edge): Pass src_idx to
16086         agg_replacements_to_vector.  Pass src_idx insstead of index to
16087         intersect_with_agg_replacements.
16089 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16091         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16092         instead of hard-wired defaults.
16094 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16096         * doc/invoke.texi (MIPS Options): Update documentation of the
16097         floating-point multiply-accumulate instruction restrictions.
16099 2013-02-21  Kostya Serebryany  <kcc@google.com>
16101         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16102         asan_shadow_offset on x86_64 linux.
16104 2013-02-21  Richard Biener  <rguenther@suse.de>
16106         PR tree-optimization/56415
16107         Revert
16108         2013-02-11  Richard Biener  <rguenther@suse.de>
16110         PR tree-optimization/56273
16111         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16112         first VRP run.
16114 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16116         PR bootstrap/56258
16117         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16118         instead of @itemx.
16120         PR inline-asm/56405
16121         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16122         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16124 2013-02-20  Jan Hubicka  <jh@suse.cz>
16126         PR tree-optimization/56265
16127         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16128         when target is referenced for first time.
16130 2013-02-20  Richard Biener  <rguenther@suse.de>
16132         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16133         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16134         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16135         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16136         not return anything.
16137         (rename_ssa_copies): Do not remove unused locals.
16138         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16139         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16140         * passes.c (execute_function_todo): Do not schedule unused locals
16141         removal if cleanup_tree_cfg did something.
16142         * tree-ssa-live.c (remove_unused_locals): Dump statistics
16143         about the number of removed locals.
16145 2013-02-20  Richard Biener  <rguenther@suse.de>
16147         PR tree-optimization/56398
16148         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16150 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16152         PR tree-optimization/55334
16153         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16154         restricted pointers to arrays.
16156 2013-02-20  Richard Biener  <rguenther@suse.de>
16157             Jakub Jelinek  <jakub@redhat.com>
16159         PR tree-optimization/56396
16160         * tree-ssa-ccp.c (n_const_val): New static variable.
16161         (get_value): Return NULL for SSA names we don't have a lattice
16162         entry for.
16163         (ccp_initialize): Initialize n_const_val.
16164         * tree-ssa-copy.c (n_copy_of): New static variable.
16165         (init_copy_prop): Initialize n_copy_of.
16166         (get_value): Return NULL_TREE for SSA names we don't have a
16167         lattice entry for.
16169 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16171         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16173 2013-02-20  Richard Biener  <rguenther@suse.de>
16175         * genpreds.c (write_lookup_constraint): Do not compare first
16176         letter of the constraint again.
16178 2013-02-20  Richard Biener  <rguenther@suse.de>
16180         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16181         and ceil_log2.
16182         (get_use_iv_cost): Terminate hashtable walk when coming across
16183         an empty entry.
16185 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
16187         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16188         reassociation for avx2 targets.
16190 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16192         * config/microblaze/microblaze.c: microblaze_has_clz = 0
16193         Add version check for v8.10.a to enable microblaze_has_clz
16194         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16195         version and TARGET_PATTERN_COMPARE check
16196         * config/microblaze/microblaze.md: New clzsi2 instruction
16198 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16200         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16201         function before branching.
16203 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16205         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16206         DUMP_INSN_RTX_UID.
16207         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16209 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16211         PR middle-end/55889
16212         * sel-sched.c: Include ira.h.
16213         (implicit_clobber_conflict_p): New function.
16214         (moveup_expr): Use it.
16215         * Makefile.in (sel-sched.o): Depend on ira.h.
16217 2013-02-19  Richard Biener  <rguenther@suse.de>
16219         PR tree-optimization/56384
16220         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16221         (vn_hash_type): Split out from ...
16222         (vn_hash_constant_with_type): ... here.
16223         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16224         (vn_phi_eq): Compare types from vn_phi_s structure.
16225         (vn_phi_lookup): Populate vn_phi_s type.
16226         (vn_phi_insert): Likewise.
16228 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
16230         PR tree-optimization/56350
16231         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16232         if haven't found reduction or nested cycle operand, rather than
16233         asserting we must find it.
16235         PR tree-optimization/56381
16236         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16237         to fold_build3.
16239 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
16240             Jakub Jelinek  <jakub@redhat.com>
16242         PR target/52555
16243         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16244         (swap_optab_enable): Same.
16245         (init_all_optabs): Use argument instead of global.
16246         * tree.h (struct tree_optimization_option): New field target_optabs.
16247         * expr.h (init_all_optabs): Add argument to prototype.
16248         (TREE_OPTIMIZATION_OPTABS): New.
16249         (save_optabs_if_changed): Protoize.
16250         * optabs.h: Declare this_fn_optabs.
16251         * optabs.c (save_optabs_if_changed): New.
16252         Declare this_fn_optabs.
16253         (init_optabs): Add argument to init_all_optabs() call.
16254         * function.c (invoke_set_current_function_hook): Handle per
16255         function optabs.
16256         * function.h (struct function): New field optabs.
16257         * config/mips/mips.c (mips_set_mips16_mode): Handle when
16258         optimization_current_node has changed.
16259         * target-globals.h (save_target_globals_default_opts): Protoize.
16260         * target-globals.c (save_target_globals_default_opts): New.
16262 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16264         PR target/56347
16265         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16266         registers %fr12 and %fr12R as call used.
16268         PR target/56214
16269         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16270         and HImode, require all displacements to be an integer multiple of
16271         their mode size.
16272         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16273         only allow QImode and HImode when reload is in progress and strict is
16274         true.  Likewise for symbolic addresses.  Use base14_operand to check
16275         displacements in REG+BASE addresses.
16277 2013-02-18  Richard Biener  <rguenther@suse.de>
16279         PR tree-optimization/56366
16280         * tree-vect-loop.c (get_initial_def_for_induction): Properly
16281         handle sign-conversion of outer-loop initial induction value.
16283 2013-02-18  Richard Biener  <rguenther@suse.de>
16285         PR middle-end/56349
16286         * cfghooks.c (merge_blocks): If we merge a latch into another
16287         block adjust references to it.
16288         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16289         (verify_loop_structure): Verify that a recorded latch is in fact
16290         a latch.
16292 2013-02-18  Richard Biener  <rguenther@suse.de>
16294         PR tree-optimization/56321
16295         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16296         order SSA name release and virtual operand unlinking.
16298 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16300         * config/microblaze/microblaze.md (save_stack_block): Define.
16301         (restore_stack_block): Likewise.
16303 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16305         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
16306         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
16307         * config/microblaze/microblaze.c (microblaze_option_override):
16308         Bail out early for PIC modes when target does not support PIC.
16310 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16312         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
16313         Replace with a microblaze version.
16314         (microblaze_trampoline_init): Adapt for microblaze.
16315         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
16316         microblaze.
16318 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
16319             Dodji Seketeli  <dodji@redhat.com>
16321         PR asan/56330
16322         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
16323         (instrument_mem_region_access): Do not forget to always put
16324         instrumentation of the of 'base' and 'base + len' in a "if (len !=
16325         0) statement, even for cases where either 'base' or 'base + len'
16326         are not instrumented -- because they have been previously
16327         instrumented.  Simplify the logic by putting all the statements
16328         instrument 'base + len' inside a sequence, and then insert that
16329         sequence right before the current insertion point.  Then, to
16330         instrument 'base + len', just get an iterator on that statement.
16331         And do not forget to update the pointer to iterator the function
16332         received as argument.
16334 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
16336         PR rtl-optimization/56348
16337         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
16339 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
16341         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
16342         (clean_graph_dump_file): Pass base to start_graph_dump.
16344 2013-02-14  Richard Henderson  <rth@redhat.com>
16346         PR target/55941
16347         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
16349 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
16351         * collect2-aix.h: Define F_LOADONLY.
16353 2013-02-14  Richard Biener  <rguenther@suse.de>
16355         PR lto/50494
16356         * varasm.c (output_constant_def_1): Get the decl representing
16357         the constant as argument.
16358         (output_constant_def): Wrap output_constant_def_1.
16359         (make_decl_rtl): Use output_constant_def_1 with the decl
16360         representing the constant.
16361         (build_constant_desc): Optionally re-use a decl already
16362         representing the constant.
16363         (tree_output_constant_def): Adjust.
16365 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
16367         Fix an asan crash
16368         * asan.c (instrument_builtin_call):  Really put the length of the
16369         second source argument into src1_len.
16371 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16373         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
16374         argument.  If it is false, don't create edge from then_bb to
16375         fallthru_bb.
16376         (insert_if_then_before_iter): Pass true to it.
16377         (build_check_stmt): Pass false to it.
16378         (transform_statements): Flush hash table only on extended basic
16379         block boundaries, rather than at the beginning of every bb.
16380         Don't flush hash table on nonfreeing_call_p calls.
16381         * tree-flow.h (nonfreeing_call_p): New prototype.
16382         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
16384 2013-02-13  David S. Miller  <davem@davemloft.net>
16386         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
16388 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
16390         PR target/56184
16391         * ira.c (max_regno_before_ira): Move from ...
16392         (ira): ... here.
16393         (fix_reg_equiv_init): Use max_regno_before_ira instead of
16394         vec_safe_length.
16396 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16398         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
16400 2013-02-13  Richard Biener  <rguenther@suse.de>
16402         PR lto/56295
16403         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
16404         globals in MEM_REFs.
16406 2013-02-13  Richard Biener  <rguenther@suse.de>
16408         * loop-init.c (loop_optimizer_init): Clear loop state when
16409         re-initializing preserved loops.
16410         * loop-unswitch.c (unswitch_single_loop): Return whether
16411         we unswitched the loop.  Do not verify loop state here.
16412         (unswitch_loops): When we unswitched a loop discover new loops.
16414 2013-02-13  Kostya Serebryany  <kcc@google.com>
16416         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
16417         on x86_64 linux.
16418         * sanitizer.def: Rename __asan_init to __asan_init_v1.
16420 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
16422         Avoid instrumenting duplicated memory access in the same basic block
16423         * Makefile.in (asan.o): Add new dependency on hash-table.h
16424         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
16425         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
16426         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
16427         (free_mem_ref_resources, has_mem_ref_been_instrumented)
16428         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
16429         (get_mem_ref_of_assignment): New functions.
16430         (get_mem_refs_of_builtin_call): Extract from
16431         instrument_builtin_call and tweak a little bit to make it fit with
16432         the new signature.
16433         (instrument_builtin_call): Use the new
16434         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
16435         of is_gimple_builtin_call.
16436         (instrument_derefs, instrument_mem_region_access): Insert the
16437         instrumented memory reference into the hash table.
16438         (maybe_instrument_assignment): Renamed instrument_assignment into
16439         this, and change it to advance the iterator when instrumentation
16440         actually happened and return true in that case.  This makes it
16441         homogeneous with maybe_instrument_assignment, and thus give a
16442         chance to callers to be more 'regular'.
16443         (transform_statements): Clear the memory reference hash table
16444         whenever we enter a new BB, when we cross a function call, or when
16445         we are done transforming statements.  Use
16446         maybe_instrument_assignment instead of instrumentation.  No more
16447         need to special case maybe_instrument_assignment and advance the
16448         iterator after calling it; it's now handled just like
16449         maybe_instrument_call.  Update comment.
16451 2013-02-13  Richard Biener  <rguenther@suse.de>
16453         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
16454         Fix loop discovery code.
16456 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
16458         PR inline-asm/56148
16459         * lra-constraints.c (process_alt_operands): Match early clobber
16460         operand with itself.  Check conflicts with earlyclobber only if
16461         the operand is not reloaded.  Prefer to reload conflicting operand
16462         if earlyclobber and matching operands are the same.
16464 2013-02-12  Richard Biener  <rguenther@suse.de>
16466         PR lto/56297
16467         * lto-streamer-out.c (write_symbol): Do not output symbols
16468         for hard register variables.
16470 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
16472         PR target/54222
16473         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
16474         (umulsidi3_insn, mulsidi3_insn): New insns.
16476 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
16478         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
16479         (struct tune_params): Add vec_costs field.
16480         * config/arm/arm.c (arm_builtin_vectorization_cost)
16481         (arm_add_stmt_cost): New functions.
16482         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
16483         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
16484         (arm_default_vec_cost): New struct of type cpu_vec_costs.
16485         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
16486         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
16487         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
16488         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
16490 2013-02-12  Richard Biener  <rguenther@suse.de>
16492         PR lto/56295
16493         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
16494         decls again if possible.
16496 2013-02-12  Richard Biener  <rguenther@suse.de>
16498         PR middle-end/56288
16499         * tree-ssa.c (verify_ssa_name): Fix check, move
16500         SSA_NAME_IN_FREE_LIST check up.
16502 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
16503             Steven Bosscher   <steven@gcc.gnu.org>
16505         PR rtl-optimization/56151
16506         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
16507         equal to op0 or op1, and last_insn pattern is CODE operation
16508         with MEM dest and one of the operands matches that MEM.
16510 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
16512         * doc/extend.texi: Document Function Multiversioning and "default"
16513         parameter string to target attribute.
16514         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
16515         target attribute parameter is "default".
16516         (ix86_compare_version_priority): Remove checks for target attribute.
16517         (ix86_mangle_function_version_assembler_name): Change error to sorry.
16518         Remove check for target attribute equal to NULL. Add assert.
16519         (ix86_generate_version_dispatcher_body): Change error to sorry.
16521 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
16522             Jack Howarth  <howarth@bromo.med.uc.edu>
16523             Patrick Marlier  <patrick.marlier@gmail.com>
16525         PR libitm/55693
16526         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
16527         define ENDFILE_SPEC as TM_DESTRUCTOR.
16528         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
16530 2013-02-11  Alexander Potapenko  <glider@google.com>
16531             Jack Howarth  <howarth@bromo.med.uc.edu>
16532             Jakub Jelinek  <jakub@redhat.com>
16534         PR sanitizer/55617
16535         * config/darwin.c (cdtor_record): Rename ctor_record.
16536         (sort_cdtor_records): Rename sort_ctor_records.
16537         (finalize_dtors): New routine to sort destructors by
16538         priority before use in assemble_integer.
16539         (machopic_asm_out_destructor): Use finalize_dtors if needed.
16541 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
16543         PR rtl-optimization/56275
16544         * simplify-rtx.c (avoid_constant_pool_reference): Check that
16545         offset is non-negative and less than cmode size before
16546         calling simplify_subreg.
16548 2013-02-11  Richard Biener  <rguenther@suse.de>
16550         PR tree-optimization/56264
16551         * cfgloop.h (fix_loop_structure): Adjust prototype.
16552         * loop-init.c (fix_loop_structure): Return the number of
16553         newly discovered loops.
16554         * tree-cfgcleanup.c (repair_loop_structures): When new loops
16555         are discovered, do a full loop-closed SSA rewrite.
16557 2013-02-11  Richard Biener  <rguenther@suse.de>
16559         PR tree-optimization/56273
16560         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16561         first VRP run.
16562         (check_array_ref): Fix missing newline in dumps.
16563         (search_for_addr_array): Likewise.
16565 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
16567         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16569 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
16571         PR target/56256
16572         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16574 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
16576         PR rtl-optimization/56246
16577         * lra-constraints.c (simplify_operand_subreg): Try to reuse
16578         reload pseudo.
16579         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16580         constraints are satisfied.
16582 2013-02-08  Jeff Law  <law@redhat.com>
16584         PR debug/53948
16585         * emit-rtl.c (reg_is_parm_p): New function.
16586         * regs.h (reg_is_parm_p): New prototype.
16587         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16588         callee-clobbered registers.
16590 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
16592         PR target/56043
16593         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16594         If there is no implicit builtin declaration, just return NULL.
16596 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16598         * config/i386/sse.md (FMAMODEM): New mode iterator.
16599         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16600         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16602 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16604         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16605         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16606         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16608 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16610         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16611         (microblaze*-*-elf): Likewise.
16612         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16613         LINK_SPEC.
16614         * config/microblaze/microblaze-c.c: Add builtin defines for
16615         _LITTLE_ENDIAN and _BIG_ENDIAN.
16616         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16617         add to TARGET_DEFAULT flags.
16618         Expand ASM_SPEC and LINK_SPEC.
16619         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16620         * config/microblaze/microblaze.md: Update extendsidi2 and
16621         movdi_internal instructions to use low-order / high-order reg
16622         print_operands.
16623         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16624         options and inversemask / mask of LITTLE_ENDIAN.
16625         * config/microblaze/t-microblaze: Expand multilib options to
16626         include mlittle-endian (le) and update exceptions patterns.
16628 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
16630         PR rtl-optimization/56195
16631         * lra-constraints.c (get_reload_reg): Don't reuse regs
16632         if they have smaller mode than requested, if they have
16633         wider mode than requested, try to return a SUBREG.
16635         PR tree-optimization/56250
16636         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16637         if type is unsigned and code isn't MULT_EXPR.
16639 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16641         PR tree-optimization/56064
16642         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16643         bits according to mode.
16644         * fixed-value.h (fixed_from_double_int)
16645         (const_fixed_from_double_int): Adjust comments.
16647 2013-02-08  Richard Biener  <rguenther@suse.de>
16649         PR lto/56231
16650         * lto-streamer.h (struct data_in): Remove current_file, current_line
16651         and current_col members.
16652         * lto-streamer-out.c (lto_output_location): Stream changed bits
16653         en-block for efficiency.
16654         * lto-streamer-in.c (clear_line_info): Remove.
16655         (lto_input_location): Cache current file, line and column
16656         globally via local statics.  Read changed bits en-block.
16657         (input_function): Do not call clear_line_info.
16658         (lto_read_body): Likewise.
16659         (lto_input_toplevel_asms): Likewise.
16661 2013-02-08  Michael Matz  <matz@suse.de>
16663         PR tree-optimization/52448
16664         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16665         (nt_call_phase): New static.
16666         (add_or_mark_expr): Only mark accesses with newer phase than any
16667         call seen.
16668         (nonfreeing_call_p): New.
16669         (nt_init_block): Update nt_call_phase, mark blocks as visited.
16670         (nt_fini_block): Keep blocks marked as visited.
16671         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16673 2013-02-08  Richard Biener  <rguenther@suse.de>
16675         * ira.c (ira): Free broken dominator information.
16677 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16679         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16681 2013-02-08  Marek Polacek  <polacek@redhat.com>
16683         * cfgloop.c (verify_loop_structure): Add more checking of headers.
16685 2013-02-08  Richard Biener  <rguenther@suse.de>
16687         PR middle-end/56181
16688         * cfgloop.h (flow_loops_find): Adjust.
16689         (bb_loop_header_p): Declare.
16690         * cfgloop.c (bb_loop_header_p): New function split out from ...
16691         (flow_loops_find): ... here.  Adjust function signature,
16692         support incremental loop structure update.
16693         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
16694         * cfgloopmanip.c (fix_loop_structure): Move ...
16695         * loop-init.c (fix_loop_structure): ... here.
16696         (apply_loop_flags): Split out from ...
16697         (loop_optimizer_init): ... here.
16698         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
16699         in incremental mode, only remove dead loops here.
16701 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16703         PR target/54222
16704         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16705         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16706         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
16707         (*round<mode>3.libgcc): New insns for fixed-modes.
16708         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
16709         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
16710         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
16711         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
16712         implementations.  Define to __builtin_avr_absFX,
16713         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
16714         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
16715         __builtin_avr_countlsFX, respectively.
16716         * config/avr/avr-c.c (target.h): Include it.
16717         (enum avr_builtin_id): New enum.
16718         (avr_resolve_overloaded_builtin): New static function.
16719         (avr_register_target_pragmas): Use it to set
16720         targetm.resolve_overloaded_builtin.
16721         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
16722         tree nodes used by DEF_BUILTIN.
16723         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
16724         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
16725         <AVR_BUILTIN_xxBITS>: Same.
16727 2013-02-08  Richard Biener  <rguenther@suse.de>
16729         * cfgloop.c (verify_loop_structure): Properly handle
16730         a loop exiting to another loop header.
16731         * ira-int.h (ira_loops): Remove.
16732         * ira.c (ira_loops): Remove.
16733         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
16734         (do_reload): Use loop_optimizer_finalize.
16735         * ira-build.c (create_loop_tree_nodes): Use get_loops and
16736         number_of_loops to access the loop tree.
16737         (more_one_region_p): Likewise.
16738         (finish_loop_tree_nodes): Likewise.
16739         (rebuild_regno_allocno_maps): Likewise.
16740         (mark_loops_for_removal): Likewise.
16741         (mark_all_loops_for_removal): Likewise.
16742         (remove_unnecessary_regions): Likewise.
16743         (ira_build): Likewise.
16744         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
16746 2013-02-08  Richard Biener  <rguenther@suse.de>
16748         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
16749         * ipa-pure-const.c (analyze_function): Avoid calling
16750         mark_irreducible_loops twice.
16751         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
16753 2013-02-07  David S. Miller  <davem@davemloft.net>
16755         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
16756         on 'reg'.
16757         * var-tracking.c (vt_add_function_parameter): Test the presence of
16758         HAVE_window_save properly and do not remap argument registers when
16759         we have a leaf function.
16761 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
16763         PR bootstrap/56227
16764         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
16765         instead of "ll".
16766         * config/i386/i386.c (ix86_print_operand): Ditto.
16768 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16770         * lra-constraints.c (process_alt_operands): Fix recently added comment.
16772 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16774         PR rtl-optimization/56225
16775         * lra-constraints.c (process_alt_operands): Check that reload hard
16776         reg can hold value for strict_low_part.
16778 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16780         PR debug/56154
16781         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
16782         dwarf2out_end_function.
16783         (in_first_function_p, maybe_at_text_label_p,
16784         first_loclabel_num_not_at_text_label): New variables.
16785         (dwarf2out_var_location): In the first function find out
16786         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
16787         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
16788         functions.
16790 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
16792         PR rtl-optimization/56178
16793         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
16794         SUBREG of a register.  Tidy up related block of code.
16795         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
16796         note if the source is a register or a SUBREG of a register.
16798 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16800         PR target/56228
16801         * config/rs6000/rs6000.md (ptrm): New mode attr.
16802         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16803         call_value_indirect_aix<pttrsize>,
16804         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16805         m in constraints.
16807 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16809         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16810         if -bnortl. Convert to strcmp and strncmp.
16812 2013-02-07  Alan Modra  <amodra@gmail.com>
16814         PR target/54009
16815         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16816         addresses won't wrap when offsetting.
16817         (rs6000_secondary_reload): Provide secondary reloads needed for
16818         wrapping LO_SUM addresses.
16820 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
16822         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16823         MACH, just __MACH__.
16825 2013-02-06  Richard Biener  <rguenther@suse.de>
16827         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16828         instead of calling fix_loop_structure.
16830 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
16832         PR middle-end/56217
16833         * omp-low.c (use_pointer_for_field): Return false if
16834         lower_send_shared_vars doesn't generate any copy-out code.
16836 2013-02-06  Tom de Vries  <tom@codesourcery.com>
16838         PR rtl-optimization/56131
16839         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16840         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
16841         of the label is NULL.  Add comment.
16843 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
16845         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16847         PR sanitizer/55374
16848         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16849         (STATIC_LIBTSAN_LIBS): Likewise.
16850         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16851         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16852         is defined, don't add anything else beyond that.
16853         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16854         (LINK_COMMAND_SPEC): Use them.
16856         PR tree-optimization/56205
16857         * tree-stdarg.c (check_all_va_list_escapes): Return true if
16858         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16859         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16861 2013-02-05  Richard Biener  <rguenther@suse.de>
16863         PR tree-optimization/53342
16864         PR tree-optimization/53185
16865         * tree-vectorizer.h (vect_check_strided_load): Remove.
16866         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16867         not disallow peeling for vectorized strided loads.
16868         (vect_check_strided_load): Make static and simplify.
16869         (vect_analyze_data_refs): Adjust.
16870         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16871         correctly when vectorizing strided loads.
16873 2013-02-05  Richard Biener  <rguenther@suse.de>
16875         * doc/install.texi: Refer to ISL, not PPL.
16877 2013-02-05  Jan Hubicka  <jh@suse.cz>
16879         PR tree-optimization/55789
16880         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16882 2013-02-05  Jan Hubicka  <jh@suse.cz>
16884         PR tree-optimization/55789
16885         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16886         the dead call anyway.
16888 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
16890         PR sanitizer/55374
16891         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16893 2013-02-04  Alexander Potapenko  <glider@google.com>
16894             Jack Howarth  <howarth@bromo.med.uc.edu>
16895             Jakub Jelinek  <jakub@redhat.com>
16897         PR sanitizer/55617
16898         * config/darwin.c (sort_ctor_records): Stabilized qsort
16899         on constructor priority by using original position.
16900         (finalize_ctors): New routine to sort constructors by
16901         priority before use in assemble_integer.
16902         (machopic_asm_out_constructor): Use finalize_ctors if needed.
16904 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
16906         PR libstdc++/54314
16907         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16908         about visibility on artificial decls.
16909         * config/sol2.c (solaris_assemble_visibility): Likewise.
16911 2013-02-04  Kai Tietz  <ktietz@redhat.com>
16913         PR target/56186
16914         * config/i386/i386.c (function_value_ms_64): Add additional valtype
16915         argument and improve checking of return-argument types for 16-byte
16916         modes.
16917         (ix86_function_value_1): Add additional valtype argument on call
16918         of function_value_64.
16919         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16920         handling infunction_value_64 function.
16922 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
16924         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16926 2013-02-04  Richard Biener  <rguenther@suse.de>
16928         PR tree-optimization/56188
16929         * tree-ssa-structalias.c (label_visit): Consider case with
16930         initially non-empty points-to set.
16931         (perform_var_substitution): Dump node mapping and clean up.
16933 2013-02-04  Richard Guenther  <rguenther@suse.de>
16935         PR lto/56168
16936         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16937         node prevail as last resort.
16938         (lto_symtab_merge_decls): Remove guard on LTRANS here.
16939         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16941 2013-02-04  Richard Biener  <rguenther@suse.de>
16943         PR tree-optimization/56113
16944         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16945         Merge into ...
16946         (equiv_class_lookup_or_add): ... this.
16947         (label_visit): Adjust and fix error in previous patch.
16948         (perform_var_substitution): Adjust.
16950 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
16952         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16953         * config/sh/sh4-300.md: Likewise.
16954         * config/sh/sh4a.md: Likewise.
16955         * config/sh/constraints.md: Likewise.
16956         * config/sh/sh.md: Likewise.
16957         * config/sh/netbsd-elf.h: Likewise.
16958         * config/sh/predicates.md: Likewise.
16959         * config/sh/sh-protos.h: Likewise.
16960         * config/sh/ushmedia.h: Likewise.
16961         * config/sh/linux.h: Likewise.
16962         * config/sh/sh.c: Likewise.
16963         * config/sh/superh.h: Likewise.
16964         * config/sh/elf.h: Likewise.
16965         * config/sh/sh4.md: Likewise.
16966         * config/sh/sh.h: Likewise.
16968 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16970         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16971         constraint to match_test floating_point_store_memory_operand().
16972         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16973         (base14_operand): New.
16974         (floating_point_store_memory_operand): New.
16975         (integer_store_memory_operand): Revise to use base14_operand and
16976         reg_plus_base_memory_operand.
16977         (move_dest_operand): Allow symbolic_memory_operands.
16978         (symbolic_memory_operand): Check for LO_SOM.
16979         (symbolic_operand): Change default case to break.
16980         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16981         CONST_DOUBLE values to be reloaded by putting them into memory when
16982         the destination is a floating point register.
16983         (movdf): Remove code to handle CONST_DOUBLE.
16984         (movsf): Likewise.
16985         (reload_indf_r1): New.
16986         (reload_insf_r1): New.
16987         Consistently use "Q" and "T" constraints with integer and floating
16988         point move instructions, respectively.
16989         (movdi): Remove FAIL.
16990         Change predicate for source operand unamed DImode move from
16991         general_operand to move_src_operand.
16992         (umulsidi3): Change predicate for destination operand to
16993         register_operand.
16994         Likewise for similar unamed patterns.
16995         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16996         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16997         (hppa_legitimize_address): Simplify mask calculation.
16998         (pa_emit_move_sequence): Revised handling of secondary reloads from
16999         REG+D addresses for floating point loads and stores.  Directly handle
17000         loading CONST0_RTX (mode) to a floating point register.
17001         (pa_secondary_reload): Handle reloading DF and SFmode constant values
17002         to floating point registers.  Don't restrict secondary reloads to
17003         floating point registers to integer modes.  Revise some comments and
17004         cleanup some code.
17005         (TARGET_LEGITIMATE_ADDRESS_P): Define.
17006         (pa_legitimate_address_p): New.
17007         (pa_legitimize_reload_address): New.
17008         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17009         (STRICT_REG_OK_FOR_BASE_P): New.
17010         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
17011         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17013 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
17014             Andrew Dixie  <andrewd@gentrack.com>
17016         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17017         flag set.
17019 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
17021         * expmed.c (extract_bit_field_1): Pass the full width of the
17022         structure to get_best_reg_extraction_insn.
17024 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17026         PR target/54601
17027         * configure.ac (use_cxa_atexit): Add AIX.
17028         * configure: Regenerate.
17030         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17032 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
17034         PR debug/54793
17035         * final.c (need_profile_function): New variable.
17036         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17037         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17038         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17039         notes, targetm.asm_out.function_prologue doesn't emit anything,
17040         HAVE_prologue and profiler should be emitted before prologue,
17041         set need_profile_function instead of emitting it.
17042         (final_scan_insn): If need_profile_function, emit
17043         profile_function on the first NOTE_INSN_BASIC_BLOCK or
17044         NOTE_INSN_FUNCTION_BEG note.
17046 2013-02-01  Richard Henderson  <rth@redhat.com>
17048         * config/rs6000/rs6000.md (smulditi3): New.
17049         (umulditi3): New.
17051         * config/alpha/alpha.md (umulditi3): New.
17053 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17055         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17056         (ASM_OUTPUT_ALIGNED_LOCAL): New.
17058 2013-02-01  Richard Biener  <rguenther@suse.de>
17060         PR tree-optimization/56113
17061         * tree-ssa-structalias.c (label_visit): Reduce work for
17062         single-predecessor nodes.
17064 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17066         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17067         range isn't testing for zero.
17069 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17071         PR middle-end/56113
17072         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17074 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17075             Nick Clifton  <nickc@redhat.com>
17077         * config/v850/constraints.md (Q): Define as a memory constraint.
17078         * config/v850/predicates.md (label_ref_operand): New predicate.
17079         (e3v5_shift_operand): New predicate.
17080         (ior_operator): New predicate.
17081         * config/v850/t-v850: Add e3v5 multilib.
17082         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17083         (v850_gen_movdi): Prototype.
17084         * config/v850/v850.c: Add support for e3v5 architecture.
17085         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17086         TARGET_V850E_UP.
17087         (construct_save_jarl): Add e3v5 long JARL support.
17088         (v850_adjust_insn_length): New function.  Adjust length of call
17089         insns when using e3v5 instructions.
17090         (v850_gen_movdi): New function: Generate instructions to move a
17091         DImode value.
17092         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17093         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17094         (TARGET_USE_FPU): Enable for e3v5.
17095         (CONST_OK_FOR_W): New macro.
17096         (ADJUST_INSN_LENGTH): Define.
17097         * config/v850/v850.md (UNSPEC_LOOP): Define.
17098         (attr cpu): Add v850e3v5.
17099         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17100         (movdi): New pattern.
17101         (movdi_internal): New pattern.
17102         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17103         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17104         (cstoresf4): Likewise.
17105         (cstoredf4): Likewise.
17106         (insv): New pattern.
17107         (rotlso3_a): New pattern.
17108         (rotlsi3_b): New pattern
17109         (rotlsi3_v850e3v5): New pattern.
17110         (doloop_begin): New pattern.
17111         (fix_loop_counter): New pattern.
17112         (doloop_end): New pattern.
17113         (branch_normal): Add e3v5 long branch support.
17114         (branch_invert): Likewise.
17115         (branch_z_normal): Likewise.
17116         (branch_z_invert): Likewise.
17117         (branch_nz_normal): Likewise.
17118         (branch_nz_invert): Likewise.
17119         (call_internal_short): Add e3v5 register-indirect JARL support.
17120         (call_internal_long): Likewise.
17121         (call_value_internal_short): Likewise.
17122         (call_value_internal_long): Likewise.
17123         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17124         (mloop): New option.
17125         * config.gcc: Add support for configuring v840e3v5 target.
17126         * doc/invoke.texi: Document new v850 specific command line options.
17128 2013-01-31  Paul Koning  <ni1d@arrl.net>
17130         PR debug/55059
17131         PR debug/54508
17132         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17133         children if parent is a class.
17134         (prune_unused_types_prune): Don't add DW_AT_declaration.
17136 2013-01-31  Richard Biener  <rguenther@suse.de>
17138         PR tree-optimization/56157
17139         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17140         match up operand with SLP child.
17142 2013-01-31  Jason Merrill  <jason@redhat.com>
17144         PR debug/54410
17145         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17146         parameters the first time.
17147         (gen_scheduled_generic_parms_dies): Check completeness here.
17149 2013-01-31  Richard Biener  <rguenther@suse.de>
17151         PR middle-end/53073
17152         * common.opt (faggressive-loop-optimizations): New flag,
17153         enabled by default.
17154         * doc/invoke.texi (faggressive-loop-optimizations): Document.
17155         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17156         infer_loop_bounds_from_undefined by it.
17158 2013-01-31  Richard Biener  <rguenther@suse.de>
17160         PR tree-optimization/56150
17161         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17162         visit virtual operands.
17163         (find_uses_to_rename_bb): Likewise.
17165 2013-01-31  Richard Biener  <rguenther@suse.de>
17167         PR tree-optimization/56150
17168         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17169         mixed store non-store stmts.
17171 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
17173         PR sanitizer/55374
17174         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17175         LIBASAN_EARLY_SPEC is defined.
17176         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17177         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17178         before %o.
17179         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17181         PR c++/55742
17182         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17183         invalid args instead of ICEing on it.
17184         (ix86_valid_target_attribute_tree): Return error_mark_node if
17185         ix86_valid_target_attribute_inner_p failed.
17186         (ix86_valid_target_attribute_p): Return false only if
17187         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
17188         target("default") attribute.
17189         (sorted_attr_string): Change argument from const char * to tree,
17190         merge in all target attribute arguments rather than just one.
17191         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
17192         instead of free.  Avoid using strcat.
17193         (ix86_mangle_function_version_assembler_name): Mangle
17194         target("default") as if no target attribute is present.  Adjust
17195         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
17196         instead of xmalloc and XDELETEVEC instead of free.
17197         (ix86_function_versions): Don't return true if one of the decls
17198         doesn't have target attribute.  If they don't and one of the decls
17199         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
17200         sorted_attr_string caller.  Use XDELETEVEC instead of free.
17201         (ix86_supports_function_versions): Remove.
17202         (make_name): Fix up formatting.
17203         (make_dispatcher_decl): Remove resolver_name and its initialization.
17204         Avoid leaking memory.
17205         (is_function_default_version): Return true if there is
17206         target("default") attribute rather than no target attribute at all.
17207         (make_resolver_func): Avoid leaking memory.
17208         (ix86_generate_version_dispatcher_body): Likewise.
17209         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17210         * target.def (supports_function_versions): Remove.
17211         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17212         * doc/tm.texi: Regenerated.
17214 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
17216         PR rtl-optimization/56144
17217         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17218         for values with side effects.
17220 2013-01-30  Richard Biener  <rguenther@suse.de>
17222         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17223         (sparseset_pop): Likewise.
17224         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
17225         to be able to use quick_push in the worker loop.
17227 2013-01-30  Marek Polacek  <polacek@redhat.com>
17229         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17231 2013-01-30  Richard Biener  <rguenther@suse.de>
17233         PR lto/56147
17234         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17236 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
17238         PR tree-optimization/56064
17239         * fixed-value.c (fixed_from_double_int): New function.
17240         * fixed-value.h (fixed_from_double_int): New prototype.
17241         (const_fixed_from_double_int): New static inline function.
17242         * fold-const.c (native_interpret_fixed): New static function.
17243         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17244         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17245         (native_encode_fixed): New static function.
17246         (native_encode_expr) <FIXED_CST>: Use it.
17247         (native_interpret_int): Move double_int worker code to...
17248         * double-int.c (double_int::from_buffer): ...this new static method.
17249         * double-int.h (double_int::from_buffer): Prototype it.
17251 2013-01-30  Richard Biener  <rguenther@suse.de>
17253         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17254         New pointer-map and obstack.
17255         (init_alias_vars): Allocate pointer-map and obstack.
17256         (delete_points_to_sets): Free them.
17257         (find_what_var_points_to): Cache result.
17258         (find_what_p_points_to): Adjust for changed interface of
17259         find_what_var_points_to.
17260         (compute_points_to_sets): Likewise.
17261         (ipa_pta_execute): Likewise.
17263 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17265         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17266         * configure: Regenerate.
17267         * config.in: Regenerate.
17268         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17269         #nobits/#progbits if supported.
17271 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
17273         PR target/56121
17274         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17275         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17276         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17278 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17280         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17281         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17283 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17285         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17286         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17288 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17290         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17291         declaration.
17292         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17293         * config/arm/cortex-a7.md: New bypasses using
17294         arm_mac_accumulator_is_result.
17296 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17298         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
17299         (cortex_a7_neon_mla): Likewise.
17300         (cortex_a7_fpfmad): New reservation.
17301         (cortex_a7_fpmacs): Use ffmas and update required units.
17302         (cortex_a7_fpmuld): Update required units and latency.
17303         (cortex_a7_fpmacd): Likewise.
17304         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
17305         (cortex_a7_neon). Likewise.
17306         (bypass) Update participating units.
17308 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17310         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
17311         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
17312         from fmac to ffma.
17313         * config/arm/vfp11.md (vfp_farith): Use ffmas.
17314         (vfp_fmul): Use ffmad.
17315         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
17316         (cortex_r4_fmacd): Use ffmad.
17317         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
17318         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
17319         (cortex_a9_fmacd): Use ffmad.
17320         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
17321         (cortex_a8_vfp_macd): Use ffmad.
17322         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
17323         (cortex_a5_fpmacd): Use ffmad.
17324         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
17325         (cortex_a15_vfp_macd): Use ffmad.
17326         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
17328 2013-01-29  Jason Merrill  <jason@redhat.com>
17330         PR libstdc++/54314
17331         * varasm.c (default_assemble_visibility): Don't warn about
17332         visibility on artificial decls.
17334 2013-01-29  Richard Biener  <rguenther@suse.de>
17336         PR tree-optimization/56113
17337         * tree-ssa-structalias.c (equiv_class_lookup): Also return
17338         the bitmap leader.
17339         (label_visit): Free duplicate bitmaps and record the leader instead.
17340         (perform_var_substitution): Adjust.
17342 2013-01-29  Richard Biener  <rguenther@suse.de>
17344         PR tree-optimization/55270
17345         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
17346         the CFG, schedule loops for fixup.
17348 2013-01-29  Nick Clifton  <nickc@redhat.com>
17350         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
17351         SP_REG.
17353 2013-01-28  Leif Ekblad  <leif@rdos.net>
17355         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
17356         * config/i386/i386.h (TARGET_RDOS): New macro.
17357         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
17358         * config/i386/i386.c (ix86_option_override_internal): For 64bit
17359         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
17360         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
17361         DEFAULT_LARGE_SECTION_THRESHOLD.
17362         * config/i386/i386.md (R14_REG, R15_REG): New constants.
17363         * config/i386/rdos.h: New file.
17364         * config/i386/rdos64.h: New file.
17366 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
17368         PR other/54814
17369         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
17370         TEST_HARD_REG_BIT.
17372 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17374         PR rtl-optimization/56117
17375         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
17376         call cselib_lookup_from_insn on the MEM before calling
17377         add_insn_mem_dependence.
17379 2013-01-28  Richard Biener  <rguenther@suse.de>
17381         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
17382         to a stmt that didn't have one.
17383         (copy_phis_for_bb): Likewise for PHI arguments.
17384         (copy_debug_stmt): Likewise for debug stmts.
17386 2013-01-28  Richard Biener  <rguenther@suse.de>
17388         PR tree-optimization/56034
17389         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
17390         (partition_builtin_p): Adjust.
17391         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
17392         it is the last partition.
17393         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
17394         up the vertex for the definition.
17395         (classify_partition): Classify whether a partition is a
17396         PKIND_REDUCTION, thus has uses outside of the loop.
17397         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
17398         Merge all PKIND_REDUCTION partitions into the last partition.
17399         (tree_loop_distribution): Seed partitions from reductions as well.
17401 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17403         PR tree-optimization/56125
17404         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
17405         pow(x,c) into sqrt(x) * powi(x, n/2) or
17406         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
17407         optimizing for size.
17408         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
17409         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
17410         integer.
17412         PR tree-optimization/56094
17413         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
17414         to UNKNOWN_LOCATION while gimplifying expr.
17416 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
17418         PR target/56114
17419         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
17420         operand 0 in movabs insn template for -masm=intel asm alternative.
17421         (*movabs<mode>_2): Ditto for operand 1.
17423 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
17425         PR target/54663
17426         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
17427         of microblaze-c.o
17429 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17431         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
17432         tm_file.
17434 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17436         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
17437         Undef to avoid warning.
17439 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17441         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
17442         * configure: Regenerate.
17444 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
17446         PR tree-optimization/56098
17447         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
17448         for stmts with volatile ops.
17449         (cond_store_replacement): Don't optimize if assign has volatile ops.
17450         (cond_if_else_store_replacement_1): Don't optimize if either
17451         then_assign or else_assign have volatile ops.
17452         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
17453         volatile ops.
17455 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17457         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
17459 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17461         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
17462         missing ':' in asm example.
17464 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
17466         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
17467         entries into lane and laneq entries.
17468         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
17469         Remove AdvSIMD scalar modes.
17470         (aarch64_sq<r>dmulh_laneq<mode>): New.
17471         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
17472         modes.
17473         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
17474         builtin implementations to relfect changes in RTL in aarch64-simd.md.
17475         * config/aarch64/iterators.md (VCOND): New.
17476         (VCONQ): New.
17478 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17480         PR target/54222
17481         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
17482         Add NULL LIBNAME argument to existing definitions.
17483         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
17484         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
17485         * config/avr/avr.c (DEF_BUILTIN): Same.
17486         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
17487         (avr_expand_builtin): Expand to a vanilla call if a libgcc
17488         implementation is available (DECL_ASSEMBLER_NAME is set).
17489         (avr_fold_absfx): New static function.
17490         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
17491         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
17492         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
17493         AVR_BUILTIN_ABSLLK.
17494         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
17495         (abshk, absk, abslk, absllk): Provide as static inline functions.
17497 2013-01-25  Marek Polacek  <polacek@redhat.com>
17499         PR tree-optimization/56035
17500         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
17502 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
17504         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
17505         (*movtf_internal_rex64): Add (!o,C) alternative
17506         (*movxf_internal_rex64): Ditto.
17507         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
17509 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
17511         * doc/invoke.texi: fix typo.
17512         * doc/objc.texi: fix typo.
17514 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
17516         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
17517         for the first two alternatives.
17519 2013-01-24  Diego Novillo  <dnovillo@google.com>
17521         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
17522         (ggc-zone.o): Remove.
17523         * configure.ac: Remove option --with-gc.
17524         * configure: Re-generate.
17525         * doc/install.texi: Remove documentation for --with-gc.
17526         * gengtype.c (write_enum_defn): Remove.  Update all users.
17527         (write_Types_process_field): Remove generation of gt_e_* argument.
17528         (output_type_enum): Remove.  Update all users.
17529         (write_enum_defn): Remove.  Update all users.
17530         (enum alloc_zone): Remove.  Update all users.
17531         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
17532         * ggc-common.c (ggc_splay_alloc): Remove first argument.
17533         Update all callers.
17534         (struct ptr_data): Remove field TYPE.  Update all users.
17535         (gt_pch_note_object): Remove argument TYPE.  Update all users.
17536         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
17537         Update all users.
17538         * ggc-none.c (ggc_alloc_typed_stat): Remove.
17539         (struct alloc_zone): Remove.
17540         (ggc_internal_alloc_zone_stat): Remove.
17541         (ggc_internal_cleared_alloc_zone_stat): Remove.
17542         * ggc-page.c (ggc_alloc_typed_stat): Remove.
17543         (ggc_pch_count_object): Remove last argument.  Update all users.
17544         (ggc_pch_alloc_object): Remove last argument.  Update all users.
17545         (struct alloc_zone): Remove.
17546         * ggc-zone.c: Remove.
17547         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
17548         (struct alloc_zone): Remove.
17549         (ggc_alloc_typed_stat): Remove.
17550         (ggc_alloc_typed): Remove.
17551         (ggc_splay_alloc): Remove first argument.
17552         (rtl_zone): Remove.  Update all users.
17553         (tree_zone): Remove.  Update all users.
17554         (tree_id_zone): Remove.  Update all users.
17555         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
17556         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
17557         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
17558         * tree-ssanames.c: Remove references to zone allocator in comments.
17560 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
17562         * config/avr/avr.c (avr_out_fract): Make register numbers that
17563         might be outside of source operand signed.
17565 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
17567         * config/i386/constraints.md (Yf): New constraint.
17568         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17569         of f constraint to conditionaly disable x87 register preferences.
17570         (*movdf_internal): Ditto.
17571         (*movsf_internal): Ditto.
17573 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
17575         PR inline-asm/55934
17576         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17577         that have operands with impossible constraints.
17578         Add a FIXME for a speed-up opportunity.
17579         * lra-constraints.c (process_alt_operands): Verify that a class
17580         selected from constraints on asms is valid for the operand mode.
17581         (curr_insn_transform): Remove incorrect comment.
17583 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
17585         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17586         TOC operand is a valid symbol ref in the constant pool.
17588 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17590         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17592 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
17594         PR target/54222
17595         * config/avr/stdfix.h: New file.
17596         * t-avr (stdfix-gcc.h): New rule to build it.
17597         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17599 2013-01-23  Kostya Serebryany  <kcc@google.com>
17601         * config/darwin.h: remove dependency on
17602         CoreFoundation (asan on Mac OS).
17604 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
17606         PR target/49069
17607         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17608         instead of cmpdi_operand for first comparison operand.
17609         Don't assert that comparison operands aren't both constants.
17611 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17613         * doc/install.texi (Downloading the Source): Update references to
17614         downloading separate components.
17616 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17618         * doc/extend.texi (__int128): Improve grammar.
17620 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
17622         PR target/56028
17623         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17624         alternative to (o,r).
17625         (*movdi_internal_rex64): Remove (!o,n) alternative.
17626         (DImode immediate->memory splitter): Remove.
17627         (DImode immediate->memory peephole2): Remove.
17628         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17629         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17630         alternative to (!o,*r).
17631         (*movtf_internal_sse): New pattern.
17632         (*movxf_internal_rex64): New pattern.
17633         (*movxf_internal): Disable for TARGET_64BIT.
17634         (*movdf_internal_rex64): Remove (!o,F) alternative.
17636 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
17638         PR middle-end/56074
17639         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17640         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17641         * tree-vect-loop-manip.c (find_loop_location): Also ignore
17642         stmt locations where LOCATION_LOCUS of the stmt location is
17643         UNKNOWN_LOCATION or BUILTINS_LOCATION.
17645         PR target/55686
17646         * config/i386/i386.md (UNSPEC_STOS): New.
17647         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17648         *strsetqi_1): Add UNSPEC_STOS.
17650 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
17652         PR c++/56067
17653         * doc/invoke.texi: Remove left over -Wsynth example.
17655 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17657         PR tree-optimization/56051
17658         * fold-const.c (fold_binary_loc): Don't fold
17659         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17660         a narrowing conversion, or widening conversion from signed
17661         to unsigned.
17663 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17665         PR rtl-optimization/56023
17666         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17667         dependent on debug instruction.
17669 2013-01-21  Martin Jambor  <mjambor@suse.cz>
17671         PR middle-end/56022
17672         * function.c (allocate_struct_function): Call
17673         invoke_set_current_function_hook earlier.
17675 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17677         * reload1.c (init_reload): Only initialize reload_obstack
17678         during the first call.
17680 2013-01-21  Marek Polacek  <polacek@redhat.com>
17682         * cfgloop.c (verify_loop_structure): Fix up grammar.
17684 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
17686         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17687         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17689 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17691         PR target/56058
17692         * config/arm/marvell-pj4.md: Update copyright year.
17693         Fix up use of alu to alu_reg and simple_alu_imm.
17695 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17697         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17699 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
17701         PR target/55433
17702         * lra-constraints.c (curr_insn_transform): Don't reuse original
17703         insn for secondary memory move when memory mode should be different.
17705 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17707         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
17708         atomic_storedi_1): New patterns.
17710 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17712         btver2 pipeline descriptions.
17713         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
17714         descriptions.
17715         * config/i386/i386.md (btver2_decode): New type attributes.
17716         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
17717         type attributes.
17718         * config/i386/btver2.md: New file describing btver2 pipelines.
17720 2013-01-19  Andrew Pinski  <apinski@cavium.com>
17722         PR tree-optimization/52631
17723         * tree-ssa-sccvn (visit_use): Before looking up the original
17724         statement, try looking up the simplified expression.
17726 2013-01-19  Anthony Green  <green@moxielogic.com>
17728         * config/moxie/moxie.c (moxie_expand_prologue): Set
17729         current_function_static_stack_size.
17731 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17733         PR tree-optimization/56029
17734         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
17735         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
17737 2013-01-18  Sharad Singhai  <singhai@google.com>
17739         PR tree-optimization/55995
17740         * dumpfile.c (dump_loc): Print location only if available.
17741         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
17743 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
17745         PR target/55433
17746         * lra-constraints.c (curr_insn_transform): Reuse original insn for
17747         secondary memory move.
17748         (inherit_reload_reg): Use rclass instead of cl for
17749         check_secondary_memory_needed_p.
17751 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17753         PR middle-end/56015
17754         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
17755         the case where writing real complex part of target modifies op1.
17757 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
17759         * config/aarch64/aarch64-simd.md
17760         (aarch64_vcond_internal<mode>): Handle unordered cases.
17761         * config/aarch64/iterators.md (v_cmp_result): New.
17763 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
17764             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17766         * config/arm/marvell-pj4.md: New file.
17767         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
17768         * config/arm/arm.md (generic_sched): Add marvell_pj4.
17769         (generic_vfp): Likewise.
17770         * config/arm/arm-cores.def: Add marvell-pj4.
17771         * config/arm/arm-tune.md: Regenerate.
17772         * config/arm/arm-tables.opt: Regenerate.
17773         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
17774         * doc/invoke.texi: Document marvell-pj4.
17776 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
17778         * config/aarch64/arm_neon.h: Map scalar types to standard types.
17780 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17782         PR debug/54114
17783         PR debug/54402
17784         PR debug/49888
17785         * var-tracking.c (negative_power_of_two_p): New.
17786         (global_get_addr_cache, local_get_addr_cache): New.
17787         (get_addr_from_global_cache, get_addr_from_local_cache): New.
17788         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
17789         heading comment.
17790         (vt_stack_offset_p): Remove.
17791         (vt_canon_true_dep): Always canonicalize loc's address.
17792         (clobber_overlapping_mems): Make sure we have a MEM.
17793         (local_get_addr_clear_given_value): New.
17794         (val_reset): Clear local cached entries.
17795         (compute_bb_dataflow): Create and release the local cache.
17796         Disable duplicate MEMs clobbering.
17797         (emit_notes_in_bb): Clobber MEMs likewise.
17798         (vt_emit_notes): Create and release the local cache.
17799         (vt_initialize, vt_finalize): Create and release the global
17800         cache, respectively.
17801         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17803 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17805         PR libmudflap/53359
17806         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17807         not found in the symtab.
17809 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17811         PR debug/56006
17812         PR rtl-optimization/55547
17813         PR rtl-optimization/53827
17814         PR debug/53671
17815         PR debug/49888
17816         * alias.c (offset_overlap_p): New, factored out of...
17817         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
17818         the conservative special case for symbolic constants.  Don't
17819         adjust zero sizes on alignment.
17821 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
17823         PR rtl-optimization/52573
17824         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17825         REG_UNUSED for the same register.
17827 2013-01-17  Richard Biener  <rguenther@suse.de>
17828             Marek Polacek  <polacek@redhat.com>
17830         PR rtl-optimization/55833
17831         * loop-unswitch.c (unswitch_loops): Move loop verification...
17832         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
17833         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17834         Set it to true when we're removing a loop from hierarchy tree in
17835         an irreducible region.
17836         (fix_bb_placements): Adjust caller.
17837         (fix_loop_placements): Likewise.
17839 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
17841         * config/avr/builtins.def (DEF_BUILTIN): Factor out
17842         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17843         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17844         Remove ID.  Adjust comments.
17845         * config/avr/avr-c.c (avr_builtin_name): Remove.
17846         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17847         * config/avr/avr.c (avr_tolower): New static function.
17848         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
17849         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17850         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17851         default expansion.
17853 2013-01-17  Jan Hubicka  <jh@suse.cz>
17855         PR tree-optimization/55273
17856         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17858 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
17860         PR target/55981
17861         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17862         store through atomic_store<mode>_1.
17863         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17865 2013-01-17  Martin Jambor  <mjambor@suse.cz>
17867         PR tree-optimizations/55264
17868         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17869         for virtual methods.
17870         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17871         virtual methods before inlining is over.
17872         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17873         virtual functions.
17874         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17875         non-virtual.
17877 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
17879         PR rtl-optimization/56005
17880         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17881         pending reads for prefetch.
17883 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
17885         * config/aarch64/aarch64.md
17886         (*cstoresi_neg_uxtw): New pattern.
17887         (*cmovsi_insn_uxtw): New pattern.
17888         (*<optab>si3_uxtw): New pattern.
17889         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17890         (*<optab>si3_insn_uxtw): New pattern.
17891         (*bswapsi2_uxtw): New pattern.
17893 2013-01-16  Richard Biener  <rguenther@suse.de>
17895         * tree-inline.c (tree_function_versioning): Remove set but
17896         never used variable.
17898 2013-01-16  Richard Biener  <rguenther@suse.de>
17900         PR tree-optimization/55964
17901         * tree-flow.h (rename_variables_in_loop): Remove.
17902         (rename_variables_in_bb): Likewise.
17903         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17904         (copy_loop_before): Adjust and delete update-ssa status.
17905         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17906         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
17907         (rename_variables_in_loop): Remove.
17908         (slpeel_update_phis_for_duplicate_loop): Likewise.
17909         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17910         use available cfg machinery instead of duplicating it.
17911         Update PHI nodes and perform poor-mans SSA update here.
17912         (slpeel_tree_peel_loop_to_edge): Adjust.
17914 2013-01-16  Richard Biener  <rguenther@suse.de>
17916         PR tree-optimization/54767
17917         PR tree-optimization/53465
17918         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17919         (vrp_visit_phi_node): For PHI arguments coming via backedges
17920         drop all symbolical range information.
17921         (execute_vrp): Compute backedges.
17923 2013-01-16  Richard Biener  <rguenther@suse.de>
17925         * doc/install.texi: Update CLooG and ISL requirements to
17926         0.18.0 and 0.11.1.
17928 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17930         PR target/55301
17931         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17932         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17933         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17935 2013-01-16  DJ Delorie  <dj@redhat.com>
17937         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17938         (UNSPECV_SP_SWITCH_E): New.
17939         (sp_switch_1): Change to an unspec.
17940         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
17941         replace $r15.
17943 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
17945         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17946         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17947         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17948         (expand_mem_thread_fence): Ditto.
17949         (expand_mem_signal_fence): Ditto.
17950         (expand_atomic_load): Ditto.
17951         (expand_atomic_store): Ditto.
17953 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17955         PR rtl-optimization/55547
17956         PR rtl-optimization/53827
17957         PR debug/53671
17958         PR debug/49888
17959         * alias.c (memrefs_conflict_p): Set sizes to negative after
17960         AND adjustments.
17962 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17964         PR target/55940
17965         * function.c (thread_prologue_and_epilogue_insns): Always
17966         add crtl->drap_reg to set_up_by_prologue.set, even if
17967         stack_realign_drap is false.
17969 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17971         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17972         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17973         *call): Fix indention.
17975 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17977         PR target/55876
17978         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17979         Update comment.
17981 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17983         PR rtl-optimization/55153
17984         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17986 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17988         PR tree-optimization/55920
17989         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17990         accesses as grp_to_be_debug_replaced.
17992 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17994         PR tree-optimization/55920
17995         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17996         there is non-useless type conversion needed from debug rhs to lhs,
17997         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17999 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
18000             Mikael Pettersson  <mikpe@it.uu.se>
18002         PR target/43961
18003         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18004         Thumb.
18005         (ASM_OUTPUT_CASE_LABEL): Remove.
18006         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18007         * final.c (shorten_branches): Update alignment of labels before
18008         jump tables if CASE_VECTOR_SHORTEN_MODE.
18010 2013-01-15  Richard Biener  <rguenther@suse.de>
18012         PR bootstrap/55961
18013         * system.h: Do not include gmp.h for building host tools.
18015 2013-01-15  Richard Biener  <rguenther@suse.de>
18017         PR middle-end/55882
18018         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18019         account for bitpos when computing alignment.
18021 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
18023         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18024         (ix86_target_macros_internal): Likewise.
18026         * config/i386/i386.c (m_CORE2I7): Removed.
18027         (m_CORE_HASWELL): New macro.
18028         (m_CORE_ALL): Likewise.
18029         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18030         (initial_ix86_arch_features): Likewise.
18031         (processor_target_table): Initializations for Core avx2.
18032         (cpu_names): New names "core-avx2".
18033         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18034         PROCESSOR_CORE_HASWELL.
18035         (ix86_issue_rate): New case.
18036         (ia32_multipass_dfa_lookahead): Likewise.
18037         (ix86_sched_init_global): Likewise.
18039         * config/i386/i386.h (TARGET_HASWELL): New macro.
18040         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18041         (processor_type): New PROCESSOR_HASWELL.
18043 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18045         PR tree-optimization/55955
18046         * tree-vect-loop.c (vectorizable_reduction): Give up early on
18047         *SHIFT_EXPR and *ROTATE_EXPR codes.
18049         PR tree-optimization/48766
18050         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18051         -ftrapv disable -fwrapv.
18053 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18055         PR target/55974
18056         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18057         etc. to 1 and not to __flash.
18058         Use LL suffix for __INT24_MAX__ with -mint8.
18059         Use ULL suffix for __UINT24_MAX__ with -mint8.
18061 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18063         * config/avr/avr-arch.h
18064         (struct base_arch_s): Use typedef avr_arch_t instead.
18065         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18066         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18067         * config/avr/avr.c: Same.
18068         * config/avr/avr-devices.c: Same.
18069         * config/avr/driver-avr.c: Same.
18070         * config/avr/gen-avr-mmcu-texi.c: Same.
18071         * config/avr/avr-mcus.def: Adjust comment.
18073 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18075         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18076         * config/aarch64/iterators.md (VALLDI): New.
18078 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18079             Andi Kleen  <ak@linux.intel.com>
18081         PR target/55948
18082         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18083         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18084         memmodel flag.
18086 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18088         * config/avr/avr-stdint.h: Remove trailing blanks.
18089         * config/avr/avr-log.h: Same.
18090         * config/avr/avr-arch.h: Same.
18091         * config/avr/avr-devices.c: Same.
18092         * config/avr/avr-dimode.md: Same.
18093         * config/avr/predicates.md: Same.
18094         * config/avr/avr-c.c: Same.  And fix typo.
18096         * config/avr/avr-protos.h: Same.  And:
18097         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18098         (init_cumulative_args): Rename to avr_init_cumulative_args.
18099         (expand_prologue): Rename to avr_expand_prologue.
18100         (expand_epilogue): Rename to avr_expand_epilogue.
18101         (adjust_insn_length): Rename to avr_adjust_insn_length.
18102         (notice_update_cc): Rename to avr_notice_update_cc.
18103         (final_prescan_insn): Rename to avr_final_prescan_insn.
18104         * config/avr/avr.c: Same.
18105         * config/avr/avr.h: Same.
18106         * config/avr/avr.md: Remove trailing blanks.
18107         (prologue): Use avr_expand_prologue.
18108         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18110 2013-01-14  Richard Biener  <rguenther@suse.de>
18112         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18113         verify_location, collect_subblocks): New functions.
18114         (verify_gimple_in_cfg): Verify that locations only reference
18115         BLOCKs in the functions BLOCK tree.
18117 2013-01-14  Richard Biener  <rguenther@suse.de>
18119         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18120         PHI argument.
18121         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18122         unshare reference.
18123         (insert_out_of_ssa_copy_on_edge): Likewise.
18124         (rewrite_close_phi_out_of_ssa): Likewise.
18125         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18126         debug expressions.
18127         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18128         propagated constants.
18129         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18130         can not be shared.
18132 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18134         * config/avr/avr-modes.def: Add GPL copyright notice.
18136 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
18138         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18139         MEMMODEL_MASK to determine memory model.
18140         (atomic_store<mode>): Ditto from operands[2].
18141         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18143 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
18145         PR fortran/55935
18146         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18147         (fold_gimple_assign): Don't call unshare_expr here.
18148         (fold_ctor_reference): Call unshare_expr.
18150 2013-01-13  Terry Guo  <terry.guo@arm.com>
18152         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18153         * doc/fragments.texi: Document MULTILIB_REUSE.
18154         * gcc.c (multilib_reuse): New internal spec.
18155         (set_multilib_dir): Also search multilib from multilib_reuse.
18156         * genmultilib (tmpmultilib3): Refactor code.
18157         (tmpmultilib4): Ditto.
18158         (multilib_reuse): New multilib argument.
18160 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
18162         * Makefile.in: Update copyright.
18164 2013-01-12  Tom de Vries  <tom@codesourcery.com>
18166         PR middle-end/55890
18167         * calls.c (expand_call): Check if arg_nr is valid.
18169 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
18171         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18172         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18173         documentation.  Add missing '__' in front of
18174         __builtin_ia32_packssdw256.
18176 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18178         PR target/55719
18179         * config/s390/s390.c (s390_preferred_reload_class): Do not return
18180         NO_REGS for larl operands.
18181         (s390_reload_larl_operand): Use s390_load_address instead of
18182         emit_move_insn.
18184 2013-01-11  Richard Biener  <rguenther@suse.de>
18186         * tree-cfg.c (verify_node_sharing_1): Split out from ...
18187         (verify_node_sharing): ... here.
18188         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18190 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
18192         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18193         Substitute TREECHECKING.
18194         * configure: Regenerate.
18195         * Makefile.in (TREECHECKING): New.
18197 2013-01-11  Richard Guenther  <rguenther@suse.de>
18199         PR tree-optimization/44061
18200         * tree-vrp.c (extract_range_basic): Compute zero as
18201         value-range for __builtin_constant_p of function parameters.
18203 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
18205         Update copyright years.
18207 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
18209         PR rtl-optimization/55672
18210         * lra-eliminations.c (mark_not_eliminable): Permit addition with
18211         const to be eliminable.
18213 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
18215         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18216         * configure: Regenerate.
18218 2013-01-10  Richard Biener  <rguenther@suse.de>
18220         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18222 2013-01-10  Richard Biener  <rguenther@suse.de>
18224         PR bootstrap/55792
18225         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18226         locations for virtual PHI arguments.
18227         (rewrite_update_phi_arguments): Likewise.
18229 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
18231         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18232         on to assembler.
18234 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
18236         PR tree-optimization/55921
18237         * tree-complex.c (expand_complex_asm): New function.
18238         (expand_complex_operations_1): Call it for GIMPLE_ASM.
18240 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18242         PR target/55718
18243         * config/s390/s390.c (s390_symref_operand_p)
18244         (s390_loadrelative_operand_p): Merge the two functions.
18245         (s390_check_qrst_address, print_operand_address): Add parameters
18246         to s390_loadrelative_operand_p invokation.
18247         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18248         (s390_reload_larl_operand, s390_secondary_reload): Use
18249         s390_loadrelative_operand_p instead of s390_symref_operand_p.
18250         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18252 2013-01-09  Mike Stump  <mikestump@comcast.net>
18254         * dse.c (record_store): Remove unnecessary assert.
18256 2013-01-09  Jan Hubicka  <jh@suse.cz>
18258         PR tree-optimization/55569
18259         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18260         * cfgloop.h (scale_loop_profile): Likewise.
18262 2013-01-09  Jan Hubicka  <jh@suse.cz>
18264         PR lto/45375
18265         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18266         functions.
18267         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18269 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
18271         PR middle-end/55114
18272         * expr.h (maybe_emit_group_store): Declare.
18273         * expr.c (maybe_emit_group_store): New function.
18274         * builtins.c (expand_builtin_int_roundingfn): Call it.
18275         (expand_builtin_int_roundingfn_2): Likewise.
18277 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
18279         PR rtl-optimization/55829
18280         * lra-constraints.c (match_reload): Add code for absent output.
18281         (curr_insn_transform): Add code for reloads of matched inputs
18282         without output.
18284 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
18286         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18287         attribute of movddup insn to DF.
18288         (*vec_interleave_lowv2df): Ditto.
18289         (vec_dupv2df): Ditto.
18291 2013-01-09  Jan Hubicka  <jh@suse.cz>
18293         PR tree-optimiation/55875
18294         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18295         EVERY_ITERATION parameter.
18296         (number_of_iterations_exit): Check if exit is executed every iteration.
18297         (idx_infer_loop_bounds): Similarly here.
18298         (n_of_executions_at_most): Simplify
18299         to only test for cases where statement is dominated by the
18300         particular bound; handle correctly the "postdominance" test.
18301         (scev_probably_wraps_p): Use max loop iterations info
18302         as a global bound first.
18304 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
18305             Nick Clifton  <nickc@redhat.com>
18307         * config/v850/v850.md (cbranchsf4): New pattern.
18308         (cstoresf4): New pattern.
18309         (cbranchdf4): New pattern.
18310         (cstoredf4): New pattern.
18311         (movsicc): Disallow floating point comparisons.
18312         (cmpsf_le_insn): Fix order of operators.
18313         (cmpsf_lt_insn): Likewise.
18314         (cmpsf_eq_insn): Likewise.
18315         (cmpdf_le_insn): Likewise.
18316         (cmpdf_lt_insn): Likewise.
18317         (cmpdf_eq_insn): Likewise.
18318         (cmpsf_ge_insn): Use LE comparison.
18319         (cmpdf_ge_insn): Likewise.
18320         (cmpsf_gt_insn): Use LT comparison.
18321         (cmpdf_gt_insn): Likewise.
18322         (cmpsf_ne_insn): Delete pattern.
18323         (cmpdf_ne_insn): Delete pattern.
18324         * config/v850/v850.c (v850_gen_float_compare): Use
18325         gen_cmpdf_eq_insn for NE comparison.
18326         (v850_float_z_comparison_operator)
18327         (v850_float_nz_comparison_operator): Move from here ...
18328         * config/v850/predicates.md: ... to here.  Move GT and GE
18329         comparisons into v850_float_z_comparison_operator.
18330         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
18331         Delete prototype.
18332         (v850_float_nz_comparison_operator): Likewise.
18334 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18336         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
18337         with calls to gen_insvsi/gen_insvdi.
18339 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18341         * config/i386/i386.c (initial_ix86_tune_features): Set up
18342         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
18344 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
18345             Jakub Jelinek  <jakub@redhat.com>
18347         PR tree-optimization/48189
18348         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
18349         If nitercst is 0, don't predict the exit edge.
18351 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18353         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
18354         in asm_fprintf with reg_names.
18355         (aarch64_print_operand_address): Likewise.
18356         (aarch64_return_addr): Likewise.
18357         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
18359 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18361         * config/pa/pa.h (VAL_U6_BITS_P): Define.
18362         (INT_U6_BITS): Likewise.
18363         * config/pa/predicates.md (uint6_operand): New predicate.
18364         (shift5_operand, shift6_operand): Likewise.
18365         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
18366         arith32_operand.
18367         (lshrdi3): Use shift6_operand.
18368         (shrpsi4, shrpdi4): New insn patterns.
18369         (extzv): Delete expander.
18370         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
18371         predicates in unamed zero extract patterns.  Tighten common constraint.
18372         (extv): Delete expander.
18373         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
18374         predicates in unamed sign extract patterns.  Tighten common constraint.
18375         (insv): Delete expander.
18376         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
18377         predicates in unamed insert patterns.  Tighten common constraint.
18378         Change uint32_operand predicate to uint6_operand predicate in unamed
18379         DImode pattern to insert constant values of type 1...1xxxx.
18381 2013-01-04  Jan Hubicka  <jh@suse.cz>
18383         PR tree-optimization/55823
18384         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
18385         issue.
18387 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18388             Uros Bizjak  <ubizjak@gmail.com>
18390         PR rtl-optimization/55845
18391         * df-problems.c (can_move_insns_across): Stop scanning at
18392         volatile_insn_p source instruction or give up if
18393         across_from .. across_to range contains any volatile_insn_p
18394         instructions.
18396 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18398         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
18399         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
18400         Declare.
18401         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
18402         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
18404 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18406         PR fortran/55341
18407         * asan.c (asan_clear_shadow): New function.
18408         (asan_emit_stack_protection): Use it.
18410 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18412         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
18413         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
18414         with tab instead of space.
18416 2013-01-08  Nick Clifton  <nickc@redhat.com>
18418         * config/rl78/rl78.c (rl78_expand_prologue): Always select
18419         register bank 0 at the start of an interrupt handler.
18420         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
18421         MDBH registers.
18423 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18425         * config/aarch64/aarch64-simd.md
18426         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
18427         (aarch64_simd_bsl): Likewise.
18428         (aarch64_vcond_internal<mode>): Likewise.
18429         (vcond<mode><mode>): Likewise.
18430         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
18431         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
18433 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18435         * config/aarch64/aarch64-builtins.c
18436         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
18438 2013-01-08  Martin Jambor  <mjambor@suse.cz>
18440         PR debug/55579
18441         * tree-sra.c (analyze_access_subtree): Return true also after
18442         potentially creating a debug-only replacement.
18444 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18446         PR middle-end/55890
18447         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
18449         PR tree-optimization/54120
18450         * tree-vrp.c (range_fits_type_p): Don't allow
18451         src_precision < precision from signed vr to unsigned_p
18452         if vr->min or vr->max is negative.
18453         (simplify_float_conversion_using_ranges): Test can_float_p
18454         against CODE_FOR_nothing.
18456 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18457             Richard Biener  <rguenther@suse.de>
18459         PR middle-end/55851
18460         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
18461         types instead of just INTEGER_TYPE types.
18463 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
18465         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
18466         TF_SIZE): Define.
18468 2013-01-07  Steve Ellcey  <sellcey@mips.com>
18470         PR target/42661
18471         * config/mips/mips.opt: Change mad to mmad to match documentation.
18473 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18475         PR target/55897
18476         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
18477         .progmemx.data now.
18479 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18481         PR target/55897
18482         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
18483         (avr_addrspace_t): Add .section_name field.
18484         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
18485         array size.
18486         (avr_addrspace): Same.  Initialize .section_name.  Remove last
18487         NULL entry.  Put __memx into .progmemx.data.
18488         (progmem_section_prefix): Remove.
18489         (avr_asm_init_sections): No need to initialize progmem_section.
18490         (avr_asm_named_section): Use avr_addrspace[].section_name to get
18491         section name prefix.
18492         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
18493         retrieve the progmem section.
18494         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
18495         boundary to run over avr_addrspace[].
18496         (avr_register_target_pragmas): Ditto.
18498 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
18500         * varasm.c (output_constant_def_contents): For asan_protect_global
18501         protected strings, adjust DECL_ALIGN if needed, before testing for
18502         anchored symbols.
18503         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18504         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
18505         normal decls.
18506         (output_object_block): For asan protected decls, emit asan padding
18507         after their contents.
18508         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
18509         (asan_finish_file): Test it here instead.
18511 2013-01-07  Nick Clifton  <nickc@redhat.com>
18512             Matthias Klose  <doko@debian.org>
18513             Doug Kwan  <dougkwan@google.com>
18514             H.J. Lu  <hongjiu.lu@intel.com>
18516         PR driver/55470
18517         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
18519         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
18521         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
18523         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
18525         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
18527 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18529         PR target/54461
18530         * doc/install.texi (Cross-Compiler-Specific Options): Document
18531         --with-avrlibc.
18533 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
18535         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
18536         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
18537         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
18538         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
18539         vqmovun_high_s64): Fix source operand number and update copyright.
18541 2013-01-07  Richard Biener  <rguenther@suse.de>
18543         PR middle-end/55890
18544         * gimple.h (gimple_call_builtin_p): New overload.
18545         * gimple.c (validate_call): New function.
18546         (gimple_call_builtin_p): Likewise.
18547         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18548         Use gimple_call_builtin_p.
18549         (find_func_clobbers): Likewise.
18550         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18551         (strlen_optimize_stmt): Likewise.
18553 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18555         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18556         (vld1q_dup_*): Likewise.
18557         (vld1_*): Likewise.
18558         (vld1q_*): Likewise.
18559         (vld1_lane_*): Likewise.
18560         (vld1q_lane_*): Likewise.
18562 2013-01-07  Richard Biener  <rguenther@suse.de>
18564         * lto-streamer.h (LTO_minor_version): Bump to 2.
18566 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18568         * config/aarch64/aarch64-protos.h
18569         (aarch64_const_double_zero_rtx_p): Rename to...
18570         (aarch64_float_const_zero_rtx_p): ...this.
18571         (aarch64_float_const_representable_p): New.
18572         (aarch64_output_simd_mov_immediate): Likewise.
18573         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18574         move immediate case.
18575         * config/aarch64/aarch64.c
18576         (aarch64_const_double_zero_rtx_p): Rename to...
18577         (aarch64_float_const_zero_rtx_p): ...this.
18578         (aarch64_print_operand): Allow printing of new constants.
18579         (aarch64_valid_floating_const): New.
18580         (aarch64_legitimate_constant_p): Check for valid floating-point
18581         constants.
18582         (aarch64_simd_valid_immediate): Likewise.
18583         (aarch64_vect_float_const_representable_p): New.
18584         (aarch64_float_const_representable_p): Likewise.
18585         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18586         (aarch64_output_simd_mov_immediate): New.
18587         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18588         (*movdf_aarch64): Likewise.
18589         * config/aarch64/constraints.md (Ufc): New.
18590         (Y): call aarch64_float_const_zero_rtx.
18591         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18593 2013-01-07  Richard Biener  <rguenther@suse.de>
18595         PR tree-optimization/55888
18596         PR tree-optimization/55862
18597         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18598         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18599         not if it is contained therein.
18601 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18603         * config/avr/t-avr: Typo.
18605 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18607         PR55243
18608         * config/avr/t-avr: Don't automatically rebuild
18609         $(srcdir)/config/avr/t-multilib
18610         $(srcdir)/config/avr/avr-tables.opt
18611         $(srcdir)/doc/avr-mmcu.texi
18612         (avr-mcus): New phony target to build them on request.
18613         (s-avr-mlib, s-avr-mmcu-texi): Remove.
18614         * avr/avr-mcus.def: Adjust comments.
18616 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
18618         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18620 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18622         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18624 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18626         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18628 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
18630         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18631         to generate profiling.
18632         * config/rs6000/aix64.h (LIB_SPEC): Same.
18634 2013-01-04  Andrew Pinski  <apinski@cavium.com>
18636         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18637         New function.
18638         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18640 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
18642         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18643         unconditionally.
18644         (ix86_expand_move): Ditto.
18645         (ix86_zero_extend_to_Pmode): Ditto.
18646         (ix86_expand_call): Ditto.
18647         (ix86_expand_special_args_builtin): Ditto.
18648         (ix86_expand_builtin): Ditto.
18650 2013-01-04  Richard Biener  <rguenther@suse.de>
18652         PR tree-optimization/55862
18653         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18654         translating them through PHI nodes.
18656 2013-01-04  Martin Jambor  <mjambor@suse.cz>
18658         PR tree-optimization/55755
18659         * tree-sra.c (sra_modify_assign): Do not check that an access has no
18660         children when trying to avoid producing a VIEW_CONVERT_EXPR.
18662 2013-01-04  Marek Polacek  <polacek@redhat.com>
18664         PR middle-end/55859
18665         * opts.c (default_options_optimization): Clarify error message.
18667 2013-01-04  Richard Biener  <rguenther@suse.de>
18669         PR middle-end/55863
18670         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18671         reassociation.
18673 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18675         PR target/53789
18676         * config/pa/pa.md (movsi): Revert previous change.
18677         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18678         references.
18680 2013-01-03  Richard Henderson  <rth@redhat.com>
18682         * config/i386/i386.c (ix86_expand_move): Always assign to op1
18683         after eliminating TLS symbols.
18685 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
18687         PR bootstrap/50167
18688         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18689         * graphite-poly.c (debug_gmp_value): Likewise.
18691 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
18693         PR target/55712
18694         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18695         selected code model, define __code_mode_small__, __code_model_medium__,
18696         __code_model_large__, __code_model_32__ or __code_model_kernel__.
18697         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18698         xchg temporary register with %k.  Declare temporary register as
18699         early clobbered.
18700         [__x86_64__]: For medium and large code models, preserve %rbx register.
18702 2013-01-03  Richard Biener  <rguenther@suse.de>
18704         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18705         (dump_subscript): Adjust.
18706         (finalize_ddr_dependent): Do not dump redundant info.
18707         (analyze_siv_subscript): Adjust.
18708         (subscript_dependence_tester): Likewise.
18709         (compute_affine_dependence): Likewise.
18711 2013-01-03  Richard Biener  <rguenther@suse.de>
18713         Revert
18714         2013-01-03  Richard Biener  <rguenther@suse.de>
18716         PR tree-optimization/55857
18717         * tree-vect-stmts.c (vectorizable_load): Do not setup
18718         re-alignment for invariant loads.
18720         2013-01-02  Richard Biener  <rguenther@suse.de>
18722         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18723         invariant load do not generate a vector load from the scalar location.
18725 2013-01-03  Richard Biener  <rguenther@suse.de>
18727         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
18728         for not vectorizing.
18729         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
18730         not build INDIRECT_REFs, call get_name once only.
18731         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
18732         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
18734 2013-01-03  Richard Biener  <rguenther@suse.de>
18736         PR tree-optimization/55857
18737         * tree-vect-stmts.c (vectorizable_load): Do not setup
18738         re-alignment for invariant loads.
18740 2013-01-03  Richard Biener  <rguenther@suse.de>
18742         PR lto/55848
18743         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
18744         prefer a built-in decl.
18746 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18748         * gcc.c (process_command): Update copyright notice dates.
18749         * gcov.c (print_version): Likewise.
18750         * gcov-dump.c (print_version): Likewise.
18752         PR rtl-optimization/55838
18753         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
18754         iv0.step, iv1.step and step.
18756 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18757             Marc Glisse  <marc.glisse@inria.fr>
18759         PR tree-optimization/55832
18760         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
18761         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
18762         integer_{one,zero}_node.
18764 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18766         PR debug/54402
18767         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
18768         * var-tracking.c (reverse_op): Don't add reverse ops to
18769         VALUEs that have already
18770         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
18772 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
18774         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
18776 2013-01-02  Teresa Johnson  <tejohnson@google.com>
18778         * dumpfile.c (dump_loc): Print filename with location.
18779         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
18780         new location_t parameter to emit complete unroll message with
18781         new dump framework.
18782         (canonicalize_loop_induction_variables): Compute loops location
18783         and pass to try_unroll_loop_completely.
18784         * loop-unroll.c (report_unroll_peel): New function.
18785         (peel_loops_completely): Use new dump format with location
18786         for main dumpfile message, and invoke report_unroll_peel on success.
18787         (decide_unrolling_and_peeling): Ditto.
18788         (decide_peel_once_rolling): Remove old dumpfile message subsumed
18789         by report_unroll_peel.
18790         (decide_peel_completely): Ditto.
18791         (decide_unroll_constant_iterations): Ditto.
18792         (decide_unroll_runtime_iterations): Ditto.
18793         (decide_peel_simple): Ditto.
18794         (decide_unroll_stupid): Ditto.
18795         * cfgloop.c (get_loop_location): New function.
18796         * cfgloop.h (get_loop_location): Declare.
18798 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18800         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18801         NULL.
18803 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18805         PR middle-end/55198
18806         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18807         BLKmode objects when EXPAND_MEMORY is specified.
18809 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18811         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18812         in loop predicate.
18813         (fold_builtin_cpu): Do not share cpu model decls across statements.
18815 2013-01-02  Jason Merrill  <jason@redhat.com>
18817         PR c++/55804
18818         * tree.c (build_array_type_1): Revert earlier change.
18820 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
18822         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18823         "cortex-a57".
18824         * config/aarch64/aarch64-tune.md: Re-generate.
18826 2013-01-02  Richard Biener  <rguenther@suse.de>
18828         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18829         invariant load do not generate a vector load from the scalar location.
18831 2013-01-02  Richard Biener  <rguenther@suse.de>
18833         PR bootstrap/55784
18834         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18835         * configure: Regenerate.
18837 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18839         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18840         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18841         (expand_builtin_int_roundingfn_2): Keep the original target around
18842         for the fallback case.
18844 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18846         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18847         to be clear for sign changes.
18849 2013-01-01  Jan Hubicka  <jh@suse.cz>
18851         * ipa-inline-analysis.c: Fix formatting.
18853 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
18855         PR tree-optimization/55831
18856         * tree-vect-loop.c (get_initial_def_for_induction): Use
18857         gsi_after_labels instead of gsi_start_bb.
18859 Copyright (C) 2013 Free Software Foundation, Inc.
18861 Copying and distribution of this file, with or without modification,
18862 are permitted in any medium without royalty provided the copyright
18863 notice and this notice are preserved.